US6223755B1 - Method and apparatus for monitoring plasma processing operations - Google Patents

Abstract

The invention generally relates to various aspects of a plasma process, and more specifically the monitoring of such plasma processes. One aspect relates in at least some manner to calibrating or initializing a plasma monitoring assembly. This type of calibration may be used to address wavelength shifts, intensity shifts, or both associated with optical emissions data obtained on a plasma process. A calibration light may be directed at a window through which optical emissions data is being obtained to determine the effect, if any, that the inner surface of the window is having on the optical emissions data being obtained therethrough, the operation of the optical emissions data gathering device, or both. Another aspect relates in at least some manner to various types of evaluations which may be undertaken of a plasma process which was run, and more typically one which is currently being run, within the processing chamber. Plasma health evaluations and process identification through optical emissions analysis are included in this aspect. Yet another aspect associated with the present invention relates in at least some manner to the endpoint of a plasma process (e.g., plasma recipe, plasma clean, conditioning wafer operation) or discrete/discernible portion thereof (e.g., a plasma step of a multiple step plasma recipe). A final aspect associated with the present invention relates to how one or more of the above-noted aspects may be implemented into a semiconductor fabrication facility, such as the distribution of wafers to a wafer production system.

Description

GOVERNMENT RIGHTS

This invention was made with the United States Government support under Contract No. DE-AC04-94AL85000 awarded by the U.S. Department of Energy. The Government has certain rights in this invention.

FIELD OF THE INVENTION

The present invention generally relates to the field of plasma processes and, more particularly, to monitoring/evaluating such plasma processes.

BACKGROUND OF THE INVENTION

Plasma is used in various types of industrial-type processes in the semiconductor and printed wiring board industries, as well as in various other industries such as in the medical equipment and automotive industries. One common use of plasma is for etching away materials in an isolated or controlled environment. Various types of materials may be etched by one or more plasma compositions, including glasses, silicon or other substrate materials, organics such as photoresist, waxes, plastics, rubbers, biological agents, and vegetable matter, and metals such as copper, aluminum, titanium, tungsten, and gold. Plasma is also utilized for depositing materials such as organics and metals onto an appropriate surface by various techniques, such as via chemical vapor deposition. Sputtering operations may also utilize plasmas to generate ions which sputter away material from a source (e.g., metals, organics) and deposit these materials onto a target such as a substrate. Surface modification operations also use plasmas, including operations such as surface cleaning, surface activation, surface passivation, surface roughening, surface smoothing, micromachining, hardening, and patterning.

Plasma processing operations can have a significant effect on a company's profit margin. This is particularly true in the semiconductor and printed wiring board industries. Consider that a single semiconductor fabrication facility may have up to 200-300 processing chambers and that each processing chamber in commercial production may process at least about 15-20 wafers per hour. Further consider that an eight inch wafer which is processed in one of these chambers in some cases may be used to produce up to 1,500 semiconductor chips which are each worth at least about $125, and that each of these semiconductor chips are in effect “pre-sold.” Therefore, a single wafer which has undergone an abnormal plasma process and which is scrapped will result in lost revenues of at least about $187,500.

The particular plasma process which acts on the wafer such that a semiconductor device may be formed therefrom is commonly referred to as a plasma recipe. Some skilled in the art refer to a plasma recipe as being a combination of one or more plasma steps, each of which is executed for a fixed period of time. However, “plasma recipe” as used in relation to the present invention means a plasma processing protocol which includes one or more different and distinct plasma steps (e.g., a certain combination of certain steps). “Different and distinct” means that each plasma step produces a different, predetermined result on the product being processed (e.g., a wafer). Differences between plasma steps may be realized by changing one or more process conditions, including without limitation the composition of the plasma, the temperature and pressure in the processing chamber, DC bias, pumping speeds, and power settings. The sequence of the plasma steps, as well as the result of each plasma step, also produces a desired overall or cumulative end result for the plasma recipe.

Plasma processes may be run on wafers in a commercial production facility in the following manner. A cassette or boat which stores a plurality of wafers (e.g., 24) is provided to a location which may be accessed by a wafer handling system associated with one or more processing chambers. One wafer at a time is processed in the chamber, although some chambers may accommodate more than one wafer at a time for simultaneous plasma processing. One or more qualification wafers may be included in each cassette, and the rest are commonly referred to as production wafers. Both the qualification and production wafers are exposed to the same plasma process in the chamber. However, no semi-conductor devices are formed from a qualification wafer as qualification wafers are processed and retained solely for testing/evaluating the plasma process, whereas semiconductor devices are formed from the production wafers. Further processing operations of these now plasma processed production wafers may be required before semiconductor devices are actually formed from such production wafers.

Monitoring is employed by many plasma processes to evaluate one or more aspects of the process. One common monitoring technique associated with plasma recipes run on wafers is endpoint detection. Current endpoint detection systems attempt to identify when a single plasma step of a given plasma recipe is complete, or more specifically that point in time when the predetermined result associated with the plasma step has been produced on the product. A representative “predetermined” result is when a layer of a multi-layered wafer has been completely removed in a manner defined by a mask or the like. Although prior art systems exist for attempting to identify the endpoint of a single step of a multiple step plasma recipe, no known system is able to identify the endpoint of each step of a multiple step plasma recipe, or even any two steps of a multiple step recipe for that matter.

Having the ability to terminate a given plasma step at its endpoint or just after endpoint is reached would reduce costs in a number of ways. Obviously, the amount of gases which are used to generate the plasma may be reduced by terminating a given plasma step when it has achieved its desired result. More importantly, terminating a given plasma step at or very shortly after its endpoint has been reached prevents the wafer from being over-etched to an undesired degree. Over-etching a wafer removes more material from the wafer than desired, such as by etching away portions of the layer immediately following that which was to be etched, and may also result in the undesirable sputtering of materials onto other portions of the wafer. The resulting effect on the semiconductor device(s) formed from this wafer may reduce the quality of the semiconductor device(s), may go undetected until the semiconductor device(s) has been delivered to the customer which would not be desirable if the device(s) was defective or deficient in any way, or both. Finally, a certain degree of over-etching of a wafer may result in the wafer simply being scrapped.

Endpoint detection is desirable in theory for plasma processes. Certain deficiencies became evident as attempts were made to implement endpoint detection techniques in commercial fabrication facilities. Initially, all known endpoint detection techniques were developed by first chemically analyzing the subject plasma operation to identify a wavelength to key in on as being indicative of endpoint. Fabrication facilities typically run a multiplicity of plasma recipes. As such, these known endpoint detection techniques increase costs due to the required retention of an experienced chemist. Moreover, these techniques often do not produce the intended result - that is the wavelength which is selected by the chemist may in fact not be at all indicative of endpoint when the plasma step is actually run since it is only “theory” based. A given endpoint detection technique may also be dependent upon the processing chamber on which the technique was developed. Accurate results may not be realized when the endpoint detection technique is used on other processing chambers. Therefore, it would be desirable to have a plasma monitoring system in which the amount of chemical “pre-analysis” is reduced and which would allow the plasma monitoring system to work to an acceptable degree on multiple processing chambers (i.e., a generic plasma monitoring system which was able to identify the relevant endpoint).

Commonly used endpoint detection techniques provide no information on how the plasma process has actually proceeded or the “health” of the plasma process—only if and when an endpoint of the subject plasma step has been reached. Other monitoring techniques which are commonly used in plasma processes suffer from this same type of deficiency. Pressures, temperatures, and flow rates of the feed gases used to form the plasma are commonly monitored. Various aspects relating to the electrical system associated with the plasma are also monitored, such as the power settings being utilized since this will affect the behavior of the plasma. However, these types of monitoring operations do not necessarily provide an indication of how the plasma process is actually proceeding. All of the “hardware” settings may be correct, but still the plasma may not be performing properly for a variety of reasons (e.g., an “unhealthy” plasma). Since errors in a plasma process are typically detected by some type of post processing, destructive testing technique, multiple wafers are typically exposed to the faulty plasma process before the error is actually identified and remedied. Therefore, it would be desirable to have a plasma monitoring system which provided a more accurate indication of how the current plasma process was actually proceeding on a more “real-time” basis, and thereby allowed for a reduction in the number of wafers which are exposed to abnormal plasma processes. Moreover, it would be desirable to have a plasma monitoring system which identified the existence of an error in the plasma process at least before the next wafer is exposed to such an “abnormal” plasma process.

Other areas of the semiconductor manufacturing process can adversely impact the profit margin of the commercial fabrication facility. Often an operator will run the wrong plasma recipe and the resulting wafers will be scrapped. It would be desirable for a plasma monitoring system to readily identify the plasma recipe being run on the given chamber to avoid this type of situation. Moreover, the length of each step of a given plasma recipe is typically set for a certain amount of time which accounts for the worst case condition (i.e., such that even the “slowest” running of the plasma step will be completed in this time frame). In many cases the step will actually be completed a significant time before this maximum setting is reached, causing the problems identified in the discussion of endpoint detection. Therefore, it would be desirable to have a plasma monitoring system which was able to identify the steps of a plasma recipe as it was being run within a processing chamber, and to utilize this information in relation to the control of the plasma process (e.g., to terminate the current step, initiate the next plasma step, or both).

Plasma processing of product (e.g., wafers) within the processing chamber will likely have an effect on the interior of the processing chamber which in turn may have an adverse effect on subsequent plasma recipes which are run on product within the chamber. Certain “byproducts” of a plasma process run on product in the chamber may be deposited on one or more interior surfaces of the chamber. These deposits may have some type of adverse effect on one or more plasma recipes which are being run in the processing chamber (e.g., a processing chamber may be used to run more than one type of plasma recipe). Deposits on the interior surfaces of the processing chamber may have the following exemplary effects on the performance of the chamber: a longer period of time may be required to reach the endpoint of one or more plasma steps of the plasma recipe; endpoint of one or more plasma steps may never be reached; and a result which is different than expected of the current plasma step may be undesirably realized (i.e., an unexpected/undesirable result). Processing chambers are typically removed from the production line on a scheduled, periodic basis for a cleaning operation to address the above-noted conditions, regardless of whether the chamber is actually in condition for a cleaning and even if the chamber was ready for cleaning well before this time. It would be desirable to have a plasma monitoring system which would provide an indication of when a processing chamber should be removed from production for cleaning.

Cleaning operations which are used to address the above-noted deposits include plasma cleans of the interior of the processing chamber, wet cleans of the interior of the processing chamber, and replacement of certain components of the processing chamber which may actually be consumed by the plasma processes conducted therein and are therefore commonly referred to as “consumables”. A plasma clean addresses the above-noted deposits by running an appropriate plasma in the processing chamber typically without any product therein (e.g., no production wafers), and therefore with the chamber being in an “empty” condition. The plasma acts on these deposits in a plasma clean and reduces the thickness thereof by chemical action, mechanical action, or both. Resulting vapors and particulate matter are exhausted from the chamber during the plasma clean. It would be desirable to have a plasma monitoring system which would provide an accurate indication of both the health and endpoint of the plasma clean currently being conducted within the processing chamber.

In some cases a plasma clean alone will not adequately address the condition of the interior of the processing chamber. Another cleaning technique which may be employed, alone or in combination with a plasma clean, is commonly referred to as a “wet clean.” Various types of solvents or the like may be used in a wet clean and are manually applied by personnel. In this regard, the subject processing chamber is depressurized, the chamber is opened to gain appropriate access, and the interior surfaces of the chamber are manually wiped down such that the solvents may remove at least some of the deposits by chemical action, mechanical action, or both. It would further be desirable to have a plasma monitoring system which would provide an accurate indication of when further execution of a plasma clean of the interior of the processing chamber will be substantially ineffective such that a wet clean may be more timely initiated or eliminated altogether.

Wet cleans and plasma cleans of the interior surfaces of the processing chamber may be ineffective in addressing deposits after a certain number of plasma processes have been conducted in the chamber. Sufficient degradation of the interior surfaces of the processing chamber may necessitate that certain components of the chamber be replaced. Components of the processing chamber which are typically replaced on some type of periodic basis are the showerhead, the wafer platform, the wafer pedestal, the quartz bell jar, and the quartz bell roof.

Additional processing of the interior surfaces of the chamber is typically undertaken after a wet clean has been performed, after one or more components of a processing chamber have been replaced and prior to resuming commercial use of the chamber (e.g., the processing of wafers in the chamber for commercial purposes), and in the case of a new chamber for that matter. No product is present in the processing chamber as a plasma is introduced into the now sealed processing chamber in this type of operation which is also commonly referred to as a plasma cleaning operation. Plasma cleaning operations in this instance address the solvent residuals from the wet clean, “prep” the new components of the chamber for plasma processing of product in the chamber, or both. It would be desirable to have a plasma monitoring system which would provide an accurate indication of both the health and endpoint of the plasma cleaning operation in this type of case.

Conditioning wafers may be run through the processing chamber before running production wafers through the processing chamber after any type of cleaning of the processing chamber, after any components of the chamber have been replaced, or in the case of a new chamber which has never had any plasma processes conducted therein. An entire plasma process is typically run on one or more conditioning wafers disposed in the subject processing chamber in a conditioning wafer operation. Conditioning wafers may simply be “blanks” or may have some semiconductor device components thereon, and the running of entire plasma processes thereon may do nothing to the conditioning wafers or portions of the conditioning wafer may be etched. Nonethless, no semiconductor devices are ever formed from a conditioning wafer and no integrated circuit of any kind is ever etched onto a conditioning wafer while running the plasma recipe thereon. Instead, conditioning wafers of this type are either refurbished (e.g., material is redeposited back into those areas which were etched during the conditioning wafer operation) and re-used again as a conditioning wafer or they are scrapped. The processing of these conditioning wafers further “preps” or “seasons” the chamber and is done for the purpose of placing the chamber in a certain condition for production. No devices are currently being used to identify when the processing of the conditioning wafers has achieved its intended purpose. Therefore, it would be desirable to have a plasma monitoring system which would provide an accurate indication of when the conditioning wafer operation may be terminated, as well as the health of such an operation.

SUMMARY OF THE INVENTION

The claims in the subject patent application are directed to the fifteenth aspect discussed in the preface to the Detailed Description section of the present patent application. The other aspects addressed in the noted preface are presented in one or more other related patent applications.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a schematic view of a wafer production system;

FIG. 2 is a perspective view of one embodiment of the wafer cassette incorporated in the wafer production system of FIG. 1;

FIGS. 3A-B are top and side views, respectively, of one embodiment of the wafer handling assembly incorporated in the wafer production system of FIG. 1;

FIG. 4 is a cross-sectional view of one embodiment of a plasma processing chamber which may be incorporated in the wafer production system of FIG. 1, namely a dry etching chamber;

FIG. 5 is a schematic view of one embodiment of a gas delivery system for the processing chamber of FIG. 4;

FIG. 6 is a schematic view of one embodiment of a plasma monitoring assembly which may be incorporated in the wafer production system of FIG. 1;

FIG. 7 is a flowchart of one embodiment of the plasma monitoring module used by the plasma monitoring assembly of FIG. 6;

FIG. 8 is a spectral pattern of one embodiment of a plasma recipe which may be run on the system of FIG. 1;

FIG. 9 is a flowchart of one embodiment of a plasma spectra directory and its various subdirectories which may be used in plasma monitoring operations;

FIG. 10 is a flowchart of one embodiment of a general data management structure which may be utilized for the various subdirectories of the plasma spectra directory of FIG. 9;

FIG. 11 is a flowchart of one embodiment of how data within the general data management structure of FIG. 10 may be condensed/consolidated;

FIG. 12A is one embodiment of a data management structure which may be used for the normal spectra subdirectory of FIG. 9;

FIG. 12B is one embodiment of a data management structure which may be used for the abnormal spectra and unknown spectra subdirectories of FIG. 9;

FIG. 13 is a flowchart of one embodiment of a pattern recognition module which may be used by the current plasma process module of FIGS. 7 and 32 in the evaluation of a plasma process being run in the processing chamber of FIG. 1;

FIG. 14 is a flowchart of one embodiment of a process alert module which may be used by the current plasma process module of FIGS. 7 and 32 in the evaluation of a plasma process being run in the processing chamber of FIG. 1;

FIG. 15 is a flowchart of one embodiment of a startup module to access the current plasma process module of FIGS. 7 and 32 for the evaluation of a plasma process being run in the processing chamber of FIG. 1;

FIG. 16 is a flowchart of one embodiment of a startup subroutine which may be accessed by the startup module of FIG. 15;

FIGS. 17A-C are exemplary spectra of one type of plasma process that may be run in any of the processing chambers of FIG.1 and monitored by the current plasma process module, namely a three-step plasma recipe;

FIGS. 18A-C are exemplary spectra of another type of plasma process that may be run in any of the processing chambers of FIG.1 and monitored by the current plasma process module, namely a plasma cleaning operation without first wet cleaning the chamber at the start, at an intermediate time, and end of such a plasma cleaning operation, respectively;

FIGS. 19A-C are exemplary spectra of another type of plasma process that may be run in any of the processing chambers of FIG.1 and monitored by the current plasma process module, namely a plasma cleaning operation conducting after a wet clean of the chamber at the start, an intermediate time, and end of such a plasma conditioning operation, respectively;

FIGS. 20A-C are exemplary spectra of another type of plasma process that may be run in any of the processing chambers of FIG.1 and monitored by the current plasma process module, namely a conditioning wafer operation at the start, an intermediate time, and end of such a conditioning wafer operation, respectively;

FIG. 21 is a flowchart of one embodiment of a plasma health subroutine which may be used by the plasma health module of FIGS. 7 and 32;

FIG. 22 is a flowchart of another embodiment of a plasma health subroutine which may be used by the plasma health module of FIGS. 7 and 32;

FIG. 23 is a flowchart of one embodiment of a plasma health/process recognition subroutine which may be used by the plasma health module of FIGS. 7 and 32;

FIG. 24 is a flowchart of another embodiment of a plasma health/process recognition subroutine which may be used by the plasma health module of FIGS. 7 and 32;

FIG. 25 is a flowchart of one embodiment of a plasma health/process step recognition subroutine which may be used by the plasma health module of FIGS. 7 and 32;

FIGS. 26A-C are exemplary spectra from a “clean” processing chamber, from an “aging” processing chamber, and from a “dirty” processing chamber, respectively;

FIG. 27 is a flowchart of one embodiment of a chamber condition subroutine which may be incorporated in the chamber condition module of FIGS. 7 and 32;

FIG. 28 is a flowchart of another embodiment of a chamber condition subroutine which may be incorporated in the chamber condition module of FIGS. 7 and 32;

FIG. 29 is a flowchart of another embodiment of a chamber condition subroutine which may be incorporated in the chamber condition module of FIGS. 7 and 32;

FIGS. 30A-D are exemplary spectra from the processing chamber in a “dirty chamber” condition, at the end of a wet clean, at the end of a plasma clean, and at the end of a conditioning wafer operation, respectively;

FIG. 31 is a schematic view of another embodiment of a plasma monitoring assembly which may be incorporated in the wafer production system of FIG. 1, which includes the current plasma process module from FIG. 7 above, and which also includes a calibration module;

FIG. 32 is a flowchart of one embodiment of a plasma monitoring module which may be used by the plasma monitoring assembly of FIG. 31, as well as by the plasma monitoring assembly of FIG. 37 below;

FIG. 33 is a schematic view of one embodiment of the spectrometer assembly of FIG. 31;

FIG. 34 is one embodiment of the fiber optic cable assembly which operatively interfaces the window and plasma monitoring assembly in FIG. 31;

FIG. 35 is a schematic view of the axes of the calibration light sent and reflected by the inner and outer surfaces of the window of FIG. 31;

FIG. 36 is one embodiment a fixture assembly for interconnecting the fiber optic cable assembly of FIG. 34 with the window on the processing chamber presented in FIG. 31;

FIG. 37 is a schematic view of another embodiment of a plasma monitoring assembly which may be incorporated in the wafer production system of FIG. 1, which includes the current plasma process module from FIG. 7 above, and which also includes a calibration module;

FIG. 38 is a schematic view of the axes of the calibration light sent and reflected by the inner and outer surfaces of the window of FIG. 37;

FIG. 39 is one embodiment a fixture assembly for interconnecting the fiber optic cables with the window on the processing chamber in the configuration presented in FIG. 37;

FIG. 40 is a flowchart of one embodiment of the calibration module from FIG. 32;

FIG. 41 is a flowchart of one embodiment of a calibration subroutine which may be used by the calibration module of FIG. 40;

FIG. 42 is one embodiment of a spectra of a calibration light which may be used by the calibration module of FIG. 40;

FIG. 43 is one embodiment of a spectra of that portion of the calibration light of FIG. 42 which is reflected by the inner surface of the processing chamber window when there are at least substantially no deposits formed thereon;

FIG. 44 is a cutaway view of another embodiment of a spectrometer which may be used by the spectrometer assembly of FIG.31 and which is operatively interfaced with the calibration module of FIG. 40;

FIG. 45 is a flowchart of another embodiment of a calibration subroutine which may be used by the calibration module of FIG. 40;

FIG. 46A is one embodiment of a spectra of a calibration light which may be used by the calibration module of FIG. 40 to identify an intensity shift condition;

FIG. 46B is another embodiment of a spectra of a calibration light which may be used by the calibration module of FIG. 40 to identify an intensity shift condition;

FIG. 47A is one embodiment of a spectra of that portion of the calibration light of FIG. 46A which is reflected by the inner surface of the processing chamber window when in a degraded or aged condition;

FIG. 47B is one embodiment of a spectra of that portion of the calibration light of FIG. 46B which is reflected by the inner surface of the processing chamber window when in a degraded or aged condition;

FIG. 48 is a flowchart of another embodiment of a calibration subroutine which may be used by the calibration module of FIG. 40;

FIG. 49 is a flowchart of one embodiment of a research subroutine which may be used by the research module of FIGS. 7 and 32;

FIGS. 50A-C are exemplary plots of intensity versus time for 3 wavelengths generated by the research subroutine of FIG. 49 from one running of a plasma recipe on product in one of the chambers from FIG. 1;

FIGS. 51A-C are exemplary plots of intensity versus time for same 3 wavelengths presented in FIGS. 50A-C, but from another running of the same plasma recipe on product in the same processing chamber;

FIG. 52 is a flowchart of one embodiment of an endpoint detection subroutine which may be used by the endpoint detection module of FIGS. 7 and 32;

FIG. 53 is a flowchart of another embodiment of an endpoint detection subroutine which may be used by the endpoint detection module of FIGS. 7 and 32;

FIG. 54A is an exemplary spectra from a processing chamber at the start of a plasma process step;

FIG. 54B is an exemplary spectra which has been selected as being indicative of endpoint of the plasma process step from FIG. 54A for use as a reference by the endpoint detection subroutine of FIG. 53;

FIG. 54C is the difference between the spectra of FIGS. 54A and 54B in accordance with the endpoint detection subroutine of FIG. 53;

FIG. 55A is an exemplary spectra from a processing chamber at an intermediate time of the plasma process step presented in FIG. 54A;

FIG. 55B is the same spectra from FIG. 54B;

FIG. 55C is the difference between the spectra of FIGS. 55A and 55B in accordance with the endpoint detection subroutine of FIG. 53;

FIG. 56A is an exemplary spectra from a processing chamber at the endpoint of the plasma process step presented in FIG. 54A;

FIG. 56B is the same spectra from FIG. 54B;

FIG. 56C is the difference between the spectra of FIGS. 55A and 55B in accordance with the endpoint detection subroutine of FIG. 53;

FIG. 57 is a flowchart of another embodiment of an endpoint detection subroutine which may be used by the endpoint detection module of FIGS. 7 and 32;

FIG. 58 is an exemplary optical emissions output which would be indicative of the endpoint of a plasma process step in accordance with the endpoint detection subroutine of FIG. 57;

FIG. 59 is a flowchart of one embodiment of a wafer distribution subroutine which may be incorporated in the wafer distribution module of FIGS. 7 and 32; and

FIG. 60 is a flowchart of another embodiment of a wafer distribution subroutine which may be incorporated in the wafer distribution module of FIGS.7 and19.

DETAILED DESCRIPTIONPreface

The present invention generally relates to various aspects of a plasma process. These aspects may be grouped into four main categories. One category relates in at least some manner to a calibration or initialization procedure, associated components, or both. The first aspect through the fourth aspect presented below are within this category. Another category relates in at least some manner to various types of evaluations which may be undertaken of a plasma process which was run, and more typically one which is currently being run, within the processing chamber (e.g., plasma health evaluations, plasma process/plasma process step identification, plasma “on” determinations). The fifth aspect through the eighth aspect presented below are within this second category. Yet another category associated with the present invention relates in at least some manner to the endpoint of a plasma process (e.g., plasma clean, conditioning wafer operation) or discrete/discernible portion thereof (e.g., a plasma step of a multiple step plasma recipe). The ninth aspect through the thirteenth aspect presented below are within this third category. Finally, the fourth category associated with the present invention relates to how one or more of the above-noted aspects may be implemented into a semiconductor fabrication facility. The fourteenth aspect through the seventeenth aspect presented below are within this fourth category.

A first aspect of the present invention is embodied in a plasma processing system having calibration capabilities in relation to the monitoring of plasma processing operations. The plasma processing system includes a processing chamber having a window with an inner surface which is exposed to plasma processes conducted within the chamber and an outer surface which is isolated from such processes. A plasma generator is associated with the plasma processing system to provide the plasma for the plasma processes. Any technique and corresponding structure for forming a plasma in the chamber is appropriate for this first aspect of the present invention. A first spectrometer assembly (e.g., one or more spectrometers of any type, such as scanning-type spectrometers and solid state spectrometers) is located outside of the chamber and is operatively interconnected with the window through a first fiber optic cable assembly (e.g., one or more fiber optic cables). A calibration light source is also located outside of the chamber and operatively interconnected with the window through a second fiber optic cable assembly (e.g., one or more fiber optic cables). Ends of the first and second fiber optic cable assemblies may be disposed on, but are preferably spaced from, the outer surface of the window to establish the noted operative interconnection in the nature of receiving data transmitted through the window.

In one embodiment of the first aspect of the present invention, “calibration” involves a comparison between data relating to the calibration light which is sent to the window by the calibration light source (e.g., a pattern, intensity, or both, of the corresponding optical emissions) and data relating to a first portion of this same calibration light which is reflected by the inner surface of the window on the processing chamber (e.g., a pattern, intensity, or both, of the first portion). The inner surface of the window of the processing chamber is that portion of the window which is typically affected by plasma processes conducted within the chamber. Changes on the inner surface of the window may have an effect on any evaluation of a plasma process being conducted within the chamber if such an evaluation is based upon the transmission of optical emissions through the window. These changes may be identified through the use of the noted calibration light since changes to the inner surface of the window will affect the reflection of the calibration light by the inner surface. Therefore, the “calibration” available from this embodiment of the first aspect of the invention may be used to calibrate an “optical emissions based” plasma monitoring system by making at least one adjustment in relation to such a system based upon the above-noted comparison. The types of “adjustments” which are contemplated by this embodiment of the first aspect are addressed in relation to the third aspect of the invention presented below.

Information on the window in relation to calibration in accordance with the first aspect of the present invention preferably includes information which is specific to the inner surface of the window on the processing chamber through which optical emissions are obtained. That is, calibrations in accordance with this first aspect are preferably in relation to only the inner surface of the window and not the outer surface of the window. Steps may be undertaken such that the portion of the calibration light which is reflected by the inner surface of the window is readily available for comparison with the calibration light in the form as it is being sent to the window in another embodiment of the subject first aspect. This may be accomplished through appropriately configuring the window. For instance, at least a portion of the window, which includes that area where the calibration light impacts the window, may have a generally wedge-shaped configuration (e.g., variable window thickness). Another characterization of a window configuration in this embodiment is that at least a portion of the inner and outer surfaces of the window may be disposed in non-parallel relation. These types of configurations are particularly useful when the relevant ends of the first and second fiber optic cable assemblies are coaxially disposed or are at least disposed in parallel relation. In this regard, the ends of the first and second fiber optic cable assemblies may be disposed such that reference axes extending from the ends of these fiber optic cable assemblies will each impact the outer surface of the window at an angle other than perpendicular, but will each impact the inner surface of the window in at least substantially perpendicular fashion. As such, that portion of the calibration light which is reflected by the outer surface of the window will be directed away from the end of the first optic cable assembly and will not be “collected” thereby. However, at least part of that portion of the calibration light which is reflected by the inner surface of the window will be directed back to the end of the first fiber optic cable assembly such that it is available to the first spectrometer assembly and thereby available for the above-noted comparison.

Another way to collect that portion of the calibration light which is reflected by the inner surface of the window instead of that which is reflected by the outer surface of the window may be done with a conventional window or where the inner and outer surfaces of the window are at least substantially parallel (e.g., a uniform thickness window configuration). In this case, the ends of the first and second fiber optic cable assemblies may be laterally displaced and directed to the same general area on the window from at least generally equal but “opposite” acute angles (e.g., “pointing” toward the same general area at an angle other than perpendicular, but from generally an “opposite” direction). To illustrate this characterization, consider a reference plane which extends through the inner and outer surfaces of the window substantially perpendicularly thereto. The end of the second fiber optic cable assembly would be disposed on a first side of this reference plane, displaced therefrom, and directed toward the window and at least generally in the direction of the reference plane such that light leaving its end would impact the outer surface of the window at an angle other than perpendicular. The end of the first fiber optic cable assembly would be disposed on a second side of this same reference plane (opposite the first side), displaced therefrom, and directed toward the window and at least generally in the direction of the reference plane such that at least a portion of the calibration light which is reflected by the inner surface of the window would be “collected” by the first fiber optic cable assembly for provision to the first spectrometer assembly. The thickness of the window will define at least in part the amount by which that portion of the calibration light which is reflected by the inner surface of the window is offset from that portion of the calibration light which is reflected by the outer surface of the window, and thereby the “sensitivity” to the relative positionings between the ends of the first and second fiber optic cable assemblies to collect only light reflected by the inner surface of the window versus its outer surface.

Anti-reflective coatings may also be applied to the outer surface of the window to reduce the effects of that portion of the calibration light which is reflected by the outer surface of the window—that is such that a comparison may be made between the calibration light that is sent to the window with that portion of the calibration light which is reflected by the inner surface of the window. A window with parallel inner and outer surfaces could be used with an arrangement whereby the ends of the first and second fiber optic cable assemblies were coaxially disposed and oriented such that reference axes projecting from their respective ends impacted both the inner and outer surfaces in at least substantially perpendicular fashion. Application of an anti-reflective coating to the outer surface of the window would reduce the amount of light which is reflected from the outer surface of the window and directed back to the first fiber optic cable assembly for provision to the first spectrometer assembly in this instance. However, there would still be some reflection from the outer surface which would be provided to the first spectrometer assembly along with the desired reflection from the inner surface. Therefore, preferably such an anti-reflective coating is used in combination with the above-noted techniques which provide for some degree of separation of the calibration light reflected from the inner and outer surfaces of the window.

Another embodiment of the first aspect of the present invention relates to a fiber optic cable fixture assembly. One application of this fixture assembly is to maintain one or more of the relevant ends of the first and second fiber optic cable assemblies in a fixed positional relationship relative to the window. Another application of this fixture assembly is to detachably interconnect one or more of, and more preferably each of, the first fiber optic cable assembly, the second fiber optic cable assembly, and the window with the processing chamber (e.g., with one or more threaded fasteners). One embodiment of such a fixture assembly is particularly useful for the case where at least the ends of the first and second fiber optic cable assemblies are coaxial and where the window is configured to provide for separation of that portion of the calibration light which is reflected by the inner surface of the window from that portion of the calibration light which is reflected by the outer surface of the window (e.g., using a generally wedge-shaped configuration for the window). In this case, the fixture assembly would include a recessed region which interfaces with or projects toward at least a portion of the outer surface of the window. At least a portion of the surface of the fixture assembly which defines this recessed region would include a light absorbing material to account for that portion of the calibration light reflected by the outer surface of the window (i.e., to absorb light which is reflected by the outer surface if it impacts the body of the fixture assembly). A first port would extend through the fixture assembly and intersect the recessed region in an orientation such that a reference axis projecting from the ends of the first and second fiber optic cable assemblies each would intersect the outer surface of the window at an angle other than perpendicular, and further such that each would intersect the inner surface of the window at least at a substantially perpendicular angle. Therefore, this embodiment of the fixture assembly may be used to retain the window and first and second fiber optic cable assemblies in a fixed relative positional relationship which allows the first fiber optic cable assembly to collect only that portion of the calibration light which is reflected by the inner surface of the window—not from the outer surface of the window.

Another embodiment of the first aspect of the present invention relates to the use of at least two different types of light by the calibration light source. One of these calibration lights may include a plurality of discrete intensity peaks, while the other of these lights may be defined by a continuum of intensity or where there are no discernible peaks (e.g., a constant intensity, a continually changing intensity, or a combination of both). In addition, one of these calibration lights may be used to identify one type of condition requiring calibration (e.g., a wavelength shift associated with the optical emissions data obtained through the window) while the other may be used for another, different type of condition requiring calibration (e.g., an intensity shift associated with the optical emissions data obtained through the window, a complete filtering of some part of the optical emissions transmitted through the window). How these types of light sources may be used to identify these types of conditions is addressed below in relation to the second aspect. Any one or more of the features discussed below on the second aspect therefore may be utilized in combination with the various features addressed in this embodiment of the first aspect of the present invention as well.

A second aspect of the present invention relates to one or more “conditions” which may be identified during some type of a calibration of a plasma processing system. Each of the various embodiments of this second aspect are embodied in a plasma processing system which includes a processing chamber, some type of a plasma monitoring assembly which monitors/evaluates (in at least some manner) plasma-based processes conducted within the chamber through optical emissions data of the plasma in the chamber during the process, and a calibration assembly which is operatively interfaced with this plasma monitoring assembly.

One embodiment of the calibration assembly associated with the second aspect of the present invention calibrates the plasma monitoring assembly for one or more conditions. One of these conditions is a wavelength shift which may be experienced in relation to the optical emissions data obtained on the subject plasma process. Another of these conditions is an intensity shift which may be experienced in relation to the optical emissions data obtained on the subject plasma process. Yet another of these conditions is where certain of the optical emissions, which should be available on the subject plasma process, are being at least substantially completely filtered (e.g., blocked out) by the window. Finally, one of these conditions is where the window is having different effects on different portions of the optical emissions. This would be the case where there are differing intensity shifts or multiple dampening effects throughout the optical emissions data being obtained on the subject plasma process. Any combination of the foregoing conditions may be identified and calibrated for by the calibration assembly of the second aspect of the present invention.

The calibration assembly discussed above in relation to the first aspect of the present invention may be used to identify and calibrate the subject plasma monitoring assembly for any of the above-noted types of conditions in relation to the second aspect of the present invention. Wavelength shifts may be identified through using a calibration light having a plurality of discrete and displaced (at different wavelengths) intensity peaks. Any shifting in the wavelengths at which these peaks appear in the calibration light which is sent to the window (calibration light) in relation to that portion of the calibration light which is reflected by the inner surface of the window (reflected light) would be indicative of a wavelength shift and which could be addressed and more preferably at least substantially alleviated by calibration. Intensity shifts may also be identified with this type of light by noting how the intensity of the peaks vary between the calibration light and the reflected light. Some peaks in the reflected light may be dampened in relation to the calibration light more than others, which would indicate the existence of multiple dampening effects. Peaks which were present in the calibration light but which were absent in the reflected light would indicate that there is filtering at those wavelengths where the peaks are absent. Preferably, intensity shifts, complete filtering, and different dampening effects are identified through using a type of calibration light having a continuum of intensity which provides a more complete picture than the case where a calibration light having discrete intensity peaks is used for any of these purposes. That is, little or no information is provided on the “behavior” of the window in relation to those wavelengths which are located between the intensity peaks in the calibration light (i.e., the effect of the window on the intensity of these wavelengths), and therefore assumptions must be made. There is no need for such assumptions in the case of using a calibration light with a continuum of intensity for the above-noted purposes.

A third aspect of the present invention is directed to monitoring a plasma process through initializing a plasma monitoring assembly. The plasma monitoring assembly evaluates at least one aspect of a subject plasma process (e.g., one currently being conducted within a processing chamber) by obtaining optical emissions data through a window on the processing chamber. Optical emissions which are obtained on the subject plasma process include at least wavelengths from about 250 nanometers to about 1,000 nanometers which defines a first wavelength range, and at least at every 1 nanometer throughout this first wavelength range.

Initialization of the plasma monitoring assembly in a first embodiment of this third aspect includes directing a calibration light toward the window through which optical emissions are obtained, reflecting a first portion of the calibration light from the window, and comparing the original calibration light which was sent with this first portion. Consequently, any combination of the various features discussed above in relation to calibration in accordance with the first and second aspects may be implemented in this third aspect as well. When the comparison of the calibration light with the first portion of the reflected light yields a first result (e.g., an intensity shift(s), a wavelength shift, filtering, or any combination thereof), at least one adjustment is made in relation to the plasma monitoring assembly.

Adjustments which may be made in relation to the plasma monitoring assembly in this first embodiment of this third aspect include physical adjustments to the plasma monitoring assembly. For instance, in the case where a spectrometer assembly is used to obtain optical emissions data and includes at least one scanning type spectrometer, the grating, one or more of its mirrors, or both may be moved (e.g., pivoted) to calibrate the plasma monitoring assembly. Any calibration of the plasma monitoring assembly involving a physical adjustment of the spectrometer assembly in this manner will typically be to address a wavelength shift which is typically due to “drifting” of the spectrometer assembly, although this type of physical adjustment may be used to address wavelength shifts from other sources. Another type of adjustment which may be made in relation to the plasma monitoring assembly is a calibration of the optical emissions which are collected or obtained on the subject plasma process, or more typically data which is representative of these optical emissions. In this regard, the “adjustment” may include the implementation of a single calibration factor or multiple calibration factors in the plasma monitoring assembly. A single calibration factor is typically utilized when there is a “uniform” intensity shift across the optical emissions to be evaluated (e.g.,±“x” intensity units possibly being considered “uniform”), whereas multiple calibration factors are typically utilized when there is a different degree of dampening across the optical emissions to be evaluated. The calibration factor(s) may then be implemented to have the desired effect on the output of the spectrometer assembly. Another way to calibrate in this manner is to normalize the optical emissions to be evaluated (or data representative thereof) based upon the comparison of the calibration light and the first portion of the calibration light reflected by the window.

Initialization of the plasma monitoring assembly in a second embodiment of this third aspect includes the steps of monitoring the window on the processing chamber through which optical emissions are obtained. The second embodiment further includes the step of determining if the window is filtering out optical emissions within a first wavelength region which is contained within the first wavelength range of about 250 nanometers to about 1,000 nanometers, which again defines the optical emissions being obtained and made available for evaluation by the plasma monitoring assembly. As such, the various features discussed above in relation to the second aspect of the invention in relation to “filtering” may be included in this second embodiment of the third aspect as well. Finally, the second embodiment of the third aspect includes the step of having the plasma monitoring assembly ignore any optical emissions within any first wavelength region or that region(s) where filtering has been detected. Notification may be provided that a filtering condition has been identified. Moreover, a recommendation that the window be replaced may be issued in this situation.

The monitoring step of the second embodiment of the third aspect may include the step of directing a calibration light toward the window, reflecting a first portion of this calibration light from the window, and comparing the calibration light with this first portion. As such, any one or more of the features discussed above in relation to the first and second aspects of the present invention may be utilized by this second embodiment of the third aspect as well. The second embodiment may also include the step of making at least one adjustment in relation to the plasma monitoring assembly when certain conditions are identified by the above-noted calibration procedure. As such, any one or more of the features discussed above in relation to the first embodiment of this third aspect may also be utilized by this second embodiment of the third aspect of the present invention.

Initialization of the plasma monitoring assembly in a third embodiment of the above-noted third aspect of the present invention includes the steps of monitoring the window on the processing chamber through which optical emissions on the subject plasma process are obtained. The third embodiment further includes the step of determining if the window is having a first effect (e.g., dampening) on a first wavelength region which is contained within the first wavelength range of about 250 nanometers to about 1,000 nanometers (which defines the optical emissions being obtained and made available for evaluation by the plasma monitoring assembly), as well as a second effect (e.g., dampening) on a second wavelength region which is also contained within the first wavelength range but outside of the first wavelength region associated with the first effect. As such, the various features discussed above in relation to the second aspect of the invention in relation to identifying different dampening effects may be included in this third embodiment of the third aspect as well. Finally, the third embodiment of the third aspect includes the step of making at least one adjustment in relation to the plasma monitoring assembly if any of these first and second types of effects are identified. As such, any one or more of the features discussed above in relation to the first embodiment of the third aspect may be utilized by this third embodiment of the third aspect as well.

The monitoring step of the third embodiment of the third aspect may include the step of directing a calibration light toward the window, reflecting a first portion of this calibration light from the window, and comparing the calibration light with this first portion. As such, any one or more of the features discussed above in relation to the first and second aspects of the present invention may be utilized by this third embodiment of the third aspect as well. The third embodiment may also include the step of making at least one adjustment in relation to the plasma monitoring assembly when certain conditions are identified by the above-noted calibration procedure. As such, any one or more of the features discussed above in relation to the first embodiment of this third aspect may also be utilized by this third embodiment of the third aspect of the present invention.

A fourth aspect of the present invention relates to a method for monitoring a plasma process which includes monitoring a window on the processing chamber in which the plasma process is conducted. In this regard, a quantity of product is loaded into the processing chamber (e.g., at least one wafer), the plasma process is thereafter conducted on this product (e.g., a plasma recipe), and data on the plasma process (e.g., optical emissions of the plasma in the chamber during the process) is obtained through the window on the processing chamber. The plasma process is evaluated based upon both the data which is obtained through the processing chamber window and the monitoring of the window.

In a first embodiment of the above-noted fourth aspect, the monitoring of the window more specifically includes the step of monitoring an actual condition of the window. The condition of the window in the case of the subject second embodiment is monitored other than through data which is obtained on the plasma process. That is, the data which is obtained on the plasma process being conducted within the processing chamber is not utilized in any manner by the step of monitoring the condition of the window in this first embodiment of the fourth aspect of the present invention.

Various features may be utilized by the above-noted first embodiment of the fourth aspect of the present invention, and these features may be used alone in the above-noted first embodiment as well as in any combination. Monitoring of the condition of the window during execution of the plasma process within the chamber may be prohibited as an additional feature of the subject first embodiment. That is, the monitoring of the condition of the window and the running of the plasma process may be executed at different and non-overlapping times. Typically, the monitoring of the condition of the window will be done prior to running of the subject plasma process within the chamber to determine the effect that the inner surface of the window will have on the subject plasma process data which is obtained through the window. As such, one or more of the features discussed above in relation to the second aspect of the present invention on “identifiable conditions” (e.g., wavelength shift, intensity shift, filtering, “uniform” dampening effects (intensity), multiple dampening effects (intensity)), as well as how these conditions may be identified (e.g., sending a calibration light toward the window and comparing this light with that portion of the calibration light which is reflected by the inner surface of the window), may be implemented in this first embodiment of fourth aspect of the present invention as well. At least one adjustment may be made in relation to the plasma monitoring assembly once one or more of these “conditions” is identified. The various types of “adjustments” are addressed above in relation to the third aspect of the present invention, and any one or more of those features may be included in this first embodiment of the fourth aspect as well.

A second embodiment of the subject fourth aspect characterizes the monitoring of the window in a different manner than as discussed above in relation to the first embodiment. In this regard, the monitoring step of this second embodiment includes the steps of directing a calibration light toward the window, reflecting a first portion of this calibration light from the inner surface of the window, and comparing the calibration light as it was sent to the window with that portion of the calibration light which was reflected by the inner surface of the window. As such, one or more of the features presented above in relation to the first and second aspects of the invention may be included in this second embodiment of the fourth aspect as well. The types of conditions which may be identified through this monitoring of the processing chamber window are presented above in relation to the second aspect of the present invention, and any one or more of these features may be included in this second embodiment of the fourth aspect as well.

A fifth aspect of the present invention relates to determining when plasma exists or is “on” within a processing chamber based upon machine-based optical analysis (i.e., not by a human eye). More specifically, the fifth aspect relates to obtaining optical emissions from within the processing chamber, evaluating these optical emissions, generating a plasma in the processing chamber, and identifying when plasma exists within the processing chamber through a machine-based evaluation of the optical emissions from within the processing chamber.

Various features may be utilized by the above-noted fifth aspect of the present invention, and these features may be used alone as well as in any combination. For instance, the identification of when plasma exists within the chamber through optical analysis may implement various techniques. The time at which the plasma comes on within the chamber may be identified by determining when the optical emissions from within the processing chamber exceeds a certain predetermined output (e.g., when the intensity of the optical emissions or a certain portion thereof within the chamber exceeds a certain amount). The identification of when plasma exists through optical analysis may also be directed toward evaluating how the optical emissions change over time. For instance, when no plasma exists within the chamber, there will be no corresponding optical emissions being emitted from the chamber. Therefore, the identifying step may simply be directed toward noting any change from a “dark” condition to a “light” condition. Another way to determine when plasma exists within the chamber through an optical analysis is to determine when the optical emissions from within the chamber include at least a certain number of discrete intensity peaks, each of which has at least a certain intensity. Finally, the presence of plasma within the chamber may be identified by determining when the current optical emissions from within the chamber match at least one output recorded on a computer-readable medium which was previously obtained from the chamber at a time when plasma was known to exist within the chamber.

Another feature which may be incorporated in the subject fifth aspect relates to the processing of a product after the plasma exists within the chamber. In one embodiment, the window on the chamber may be monitored in accordance with the fourth aspect of the invention discussed above. These monitoring operations may be automatically terminated at a time when plasma is first identified within the chamber through the noted optical analysis provided by this fifth aspect. In another embodiment, plasma processes conducted within the chamber may be monitored by a plasma monitoring assembly. Calibration of this plasma monitoring assembly may be made available in accordance with the third aspect of the invention discussed above. These calibration operations may be automatically terminated when plasma is identified within the chamber through the noted optical analysis provided by this fifth aspect.

A sixth aspect of the present invention relates to a plasma spectra directory which contains at least optical emissions data from plasma processes previously conducted within the processing chamber and which are used to evaluate plasma processes subsequently conducted in this very same processing chamber. The plasma spectra directory is stored on a computer-readable storage medium and for ease of description includes a first data structure having a plurality of data entries. Each of these data entries includes data representative of optical emissions from at least one time during the subject plasma process, and this data is associated with one of a first category, a second category, and a third category.

The data entries associated with the first category are those plasma processes which have been run in the chamber and which define a “standard” against which subsequent plasma processes are judged. Plasma processes which are run in the processing chamber are evaluated to determine if they “correspond” or “match” with at least one data entry associated with the first category. These types of plasma processes associated with the first category thereby may be characterized as “normal” runs. In this case, plasma processes which are associated with the first category are at least assumed to have proceeded without substantially any error, and may be tested in some manner to confirm that they did in fact proceed without any substantial error or aberration.

The optical emissions of the plasma in the processing chamber will typically reflect whether a given plasma process is proceeding in a “normal” fashion. In this regard, optical emissions associated with a data entry of the first category preferably include at least wavelengths from about 250 nanometers to about 1,000 nanometers at least at every 1 nanometer throughout this range and at least at every 1 second from the subject plasma process. Although evaluations of plasma processes subsequently conducted in the subject processing chamber need not utilize all of this data, it will be available if desired/required. Moreover, typically data on the entirety of the subject plasma process, or at least that portion of the process after the plasma has stabilized, is included in data entries associated with the first category.

Virtually any type of plasma process may be included in data entries associated with the first category as long as its optical emissions data provides an indication that the plasma process is proceeding in a certain fashion. One or more plasma recipes (run on production wafers, qualification wafers, or both), plasma cleanings (before or after a wet clean), and conditioning wafer operations may each be included in the plasma spectra directory and associated with the first category. Multiple “species” of these types of plasma processes may also be included in the plasma spectra directory in association with the first category (e.g., different types of plasma recipes). Multiple data entries of the same “species” may also be included in the plasma spectra directory in association with the first category as well (e.g., multiple entries of the same plasma recipe run on the same type of product).

The data entries associated with the second category of the subject sixth aspect are those plasma processes (e.g., plasma recipes, plasma cleans, conditioning wafer operations) which have been run in the processing chamber and which have encountered at least one error or aberration. This error or aberration will typically be represented by a change in the optical emissions of the plasma in the processing chamber, and the cause may be identified by a review of these optical emissions. Typically this review is after termination of the subject plasma process. Obtaining optical emissions data within the above-noted wavelength range increases the likelihood that optical emissions data which is representative of the error or aberration will in fact be available for inclusion in a data entry which is associated with the second category.

The identification or cause of the error or aberration is included in some manner with the data entry which is associated with the second category. Various actions may be initiated based upon this information. An alert or the like (audio, visual, or both) may be issued to apprise personnel that an error has been encountered in the subject plasma process in the subject processing chamber. Specific information on the error may also be made available, as well as one or more ways to address or correct the error or aberration. Finally, corrective actions may be automatically undertaken if desired.

Typically, the entire run in which the error occurred is not included in the data entry associated with the second category. Instead, only those optical emissions which reflect the existence of the subject error or aberration are typically included in such a data entry. This may include optical emissions data from only a single point in time during the subject plasma process or from multiple times. Optical emissions included in any data entry associated with the second category may also be of the above-noted wavelength range. However, if the error or aberration is only reflected in a certain portion of the optical emissions which are obtained on the subject plasma process, only this portion need be included in the plasma spectra directory for the subject data entry associated with the second category.

The data entries associated with the third category in relation to the subject sixth aspect are those plasma processes which have been run in the processing chamber and which are “unknown” to the plasma spectra subdirectory. That is, the optical emissions from the subject plasma process have failed to correspond with any data entry associated with the first category or with the second category. Moreover the reason as to why this is the case has yet to be determined, or more accurately the cause has yet to be associated with a data entry on the computer-readable storage medium. Two situations will typically encompass each case where a data entry is recorded in the plasma spectra directory and associated with the third category. Plasma processes which have not yet been recorded in the plasma spectra directory and associated with the first category are one such situation. In this case, the entirety of the subject plasma process may be recorded in the plasma spectra directory and associated with the third category. Once this data entry is identified as being a new plasma process which did or was assumed to have proceeded without substantially any error or aberration, the data entry may be “transferred” from the third category to the first category. Plasma processes which have encountered an error which has not been recorded in the plasma spectra directory and associated with the second category will also be recorded in a data entry under the third category. Typically, only data from the initial occurrence of the error or aberration until the end of the plasma process will be recorded in a data entry associated with the third category in this situation. Subsequent evaluation of the optical emissions data from this plasma process may reveal that a “new” error was encountered. If the cause of the error is identified, all or a portion of the data from the subject data entry associated with the third category may then be “transferred” to the second category.

A seventh aspect of the present invention relates to various analytical techniques which may be used to evaluate a plasma process in at least some mariner. In a first embodiment of this seventh aspect, a computer-readable storage medium includes a plurality of data entries. At least one of these data entries is associated with the type of first category discussed above in relation to the sixth aspect, while at least one of these data entries is associated with the type of second category also discussed above in relation to the sixth aspect. The evaluation technique embodied by this first embodiment of the seventh aspect first determines if the subject plasma process corresponds with any data entry associated with the first category. Any such correspondence may be used to characterize the subject plasma process as “normal” or the like. If the subject plasma process at any time fails to correspond with at least one data entry under the first category, this first embodiment of the seventh aspect will then “search” those data entries under the second category to see if the subject plasma process has encountered a known error or aberration. Therefore, data entries under the second category are not searched in each case.

Various features may be utilized by the above-noted first embodiment of the seventh aspect of the present invention, and these features may be used alone in the above-noted first embodiment as well as in any combination. Initially, each of the various features/concepts discussed above in relation to the sixth aspect are equally applicable to and may be incorporated in this first embodiment of the subject seventh aspect. There are also various ways of determining whether the optical emissions data of the subject plasma process conforms or corresponds with a given data entry. Conformance or correspondence may be based upon determining if the pattern of the current optical emissions is a “match” with the pattern of the relevant optical emissions from the data entry. What are the “relevant” optical emissions may also be subject to a number of characterizations. For instance, the time associated with the current optical emissions may be used as a criterion to determine whether these emissions correspond with a given data entry. That is, the time at which the subject optical emissions were obtained would be used to identify which optical emissions from a given data entry would be used for the subject comparison (i.e., select the optical emissions from the data entry which were obtained at the same time as the subject optical emissions). Alternatively, the subject plasma process may simply be evaluated to determine if it is progressing in the same fashion as at least one of the data entries associated with the first category, although not necessarily at the same rate. Time would not be a limiting criterion in this case.

Various actions may be initiated if the current plasma process corresponds with a data entry associated with the second category, either manually or automatically. For instance, the subject plasma process may be terminated, an alert may be issued that an error has been encountered, further use of the processing chamber for processing product may be suspended, adjustment of the plasma process may be undertaken in an attempt to remedy the subject error(s), or any combination thereof.

A second embodiment of the subject seventh aspect utilizes a computer-readable storage medium which includes a first data entry which is associated with a first category of the type identified above in relation to the sixth aspect. This data entry includes a plurality of first data segments from a plurality of different times during one plasma process previously conducted in the processing chamber. Each data segment includes optical emissions of the plasma in the chamber for wavelengths of at least about 250 nanometers to about 1,000 nanometers which defines a first wavelength range, and at least at every 1 nanometer throughout this first wavelength range. This second embodiment entails obtaining current optical emissions from another plasma process run in this same processing chamber which are also within the first wavelength range and at least at every 1 nanometer throughout this first wavelength range. A comparison is undertaken between the current optical emissions and those associated with at least one first data segment of the first data entry throughout the first wavelength range and at least at every 1 nanometer throughout the first wavelength range. The features discussed in relation to this second embodiment may be incorporated with those discussed above in relation to the first embodiment of the subject seventh aspect.

A third embodiment of the subject seventh aspect also utilizes a computer-readable storage medium. A first plasma process is run in the processing chamber. Optical emissions of the plasma in the chamber are obtained for wavelengths of at least about 250 nanometers to about 1,000 nanometers which defines a first wavelength range, and at least at every 1 nanometer throughout this first wavelength range. This data is obtained at a plurality of times during this first plasma process and such is recorded in a first data entry on the computer-readable storage medium. A second plasma process is conducted after termination of the first and similar data is obtained. The second plasma process is evaluated based upon at least a portion of the optical emissions data from the second plasma process. In some cases it may be desirable to compare the optical emissions from the second plasma process with those optical emissions from the first plasma process throughout the entirety of the first wavelength range and at least at every 1 nanometer throughout this first wavelength range. However, in some cases this may not be practical, desirable, or necessary. In this regard, the progress of the second plasma process in relation to the first plasma process recorded in the first data entry on the computer-readable storage medium may be based upon an evaluation of at least a 50 nanometer bandwidth and at least every 1 nanometer throughout this smaller bandwidth.

A smaller wavelength region may be selected for evaluating the second plasma process in relation to the first plasma process in a variety of manners. The particular wavelength(s) at which error(s) have been previously encountered in running this same plasma process may be used to select that portion of the first wavelength range which should be used in the subject evaluation (e.g., ±25 nanometers of each wavelength which is indicative of an error or aberration). Moreover, a wavelength region may be selected which includes each of the errors previously encountered on the same type of plasma process. The “width” of the region may be defined by the two extreme wavelengths, although it would be preferred to include a “buffer” of sorts on each of these ends (e.g., expand the range by 25 nanometers on each end). Finally, the particular wavelength(s) which is indicative of the endpoint of the subject plasma process or discrete/discernible portion thereof may be used to select that portion of the first wavelength range which should be used in the subject evaluation (±25 nanometers of each such wavelength). Individual endpoint indicator wavelengths are discussed in more detail below in relation to the ninth aspect of the present invention.

An eighth aspect of the present invention relates to identifying the type of plasma process conducted within the processing chamber. This aspect may be used to identify whether a plasma process is a certain type of plamsa recipe being run on a certain type of production wafer, a certain type of plasma recipe being run on a certain type of qualification wafer, a certain type of plasma recipe being run on a certain type of conditioning wafer, or a plasma clean being run in a chamber. A first embodiment of this eighth aspect is able to identify the particular type of a plasma recipe being run on product (e.g., production wafer, qualification wafer) in a processing chamber based upon the storage of at least two plasma recipes on a computer-readable storage medium. In this regard, the computer-readable storage medium includes a plurality of data entries. A first of these data entries includes relevant data from a plurality of times during a first plasma recipe run on product in the processing chamber (and preferably of the entirety of this first plasma recipe at least after stabilization of the plasma). A second of these data entries includes relevant data from a plurality of times during a second plasma recipe (different from the first plasma recipe) run on product in the same processing chamber (and preferably of the entirety of this second plasma recipe at least after stabilization of the plasma). Data on a subject plasma recipe which is being run on product in the same processing chamber is obtained. This data is used to determine if the current plasma recipe is of the same type as the first or second plasma recipe stored on the computer-readable storage medium. Preferably, this determination is completed prior to termination of the current plasma recipe and at least before the next product is loaded into the chamber. This first embodiment of the eighth aspect may be used to determine not only the identity of the subject plasma process, but the type of product (e.g., whether a production wafer or a qualification wafer) that is being processed by including relevant data from prior plasma processes on the computer-readable storage medium. That is, by including a plasma recipe “A” run on a certain type of production wafer in one data entry and the same plasma recipe “A” on a certain type of qualification wafer in another data entry, the ability exists to determine if the current plasma recipe is being run on a production versus a qualification wafer.

Various features may be utilized by the above-noted first embodiment of the eighth aspect of the present invention, and these features may be used alone in the above-noted first embodiment as well as in any combination. Initially, the data obtained on the current plasma process may be optical emissions of the plasma in the processing chamber. These optical emissions may include at least wavelengths from about 250 nanometers to about 1,000 nanometers (inclusive) which defines a first wavelength range, and optical emissions may be obtained at least at every 1 nanometer throughout this first wavelength range. Optical emissions of the subject plasma process may be compared with one or both of the first and second plasma recipes stored on the computer-readable storage medium to see if there is sufficient correspondence therebetween. As such, the techniques discussed above in relation to the seventh aspect may be implemented in this first embodiment of the eighth aspect as well.

A second embodiment of the subject eighth aspect is directed toward inputting the plasma recipe to be run in the chamber and using the principles discussed above in relation to the first embodiment of the eighth aspect to verify that no errors were made when inputting the subject plasma recipe. That is, the identify of the subject plasma process is determined in accordance with the first embodiment of this eighth aspect. Therefore, each of the various features discussed above in relation to the first embodiment of the eighth aspect may be incorporated in this second embodiment of the eighth aspect as well. The identification of the subject plasma process through optical analysis in accordance with the foregoing is then conveyed in some manner to the appropriate personnel (e.g., on a display). If the wrong plasma process was input for a certain wafer or “lot” of wafers, the identification of the subject plasma process and the conveyance of this identity to the appropriate personnel would apprise the personnel of this situation.

A third embodiment of the subject eighth aspect is directed to identifying a subject plasma recipe based upon at least two plasma recipes which are stored on a computer-readable storage medium and which were previously run in the same processing chamber. The first execution of the subject plasma recipe is initiated and is of the type associated with either the first or second plasma recipe. At least one characteristic of the plasma is monitored during the execution of each subject plasma recipe. Both the first and second plasma recipes are available for comparison against the first execution of the subject plasma recipe. However, after the first execution of plasma recipe is identified as being either the first or second plasma recipe from the computer-readable storage medium (identified plasma recipe), subsequent executions of the subject plasma recipes are evaluated at least initially only in relation to the identified plasma recipe on the computer-readable storage medium. This embodiment is particularly pertinent to the case where the first wafer of a cassette or boat of wafers is evaluated in accordance with the foregoing since the same plasma recipe is typically run on the entire cassette. Therefore, once the third embodiment of the eighth aspect determines the identify of the plasma recipe being run on the first wafer, all subsequent wafers in the cassette are at least initially evaluated against only one plasma recipe on the computer-readable storage medium. Enhanced evaluation speed therefore may be realized. If any such subsequent executions of the subject plasma recipe deviate from the identified plasma recipe on the computer-readable storage medium, one variation of this third embodiments makes other plasma recipes on the computer-readable storage medium available for evaluation of the then current plasma recipe. Another possibility would be to check for any data entry of the same plasma recipe that was run on a qualification wafer versus a production wafer if there was a failure of the current plasma recipe to correspond with the plasma recipe stored on the computer-readable storage medium (i.e., assuming that the first wafer was a production wafer and had its plasma recipe identified). In this case, the logic would be to evaluate the entire cassette first against the plasma recipe for the production wafer stored on the computer-readable storage medium, and then against the same plasma recipe but for a qualification wafer stored on the computer-readable storage medium if necessary.

A ninth aspect of the present invention relates to engaging in research to identify one or more indicators of a first endpoint which is when the plasma process (e.g., plasma recipe, plasma clean, conditioning wafer operation) or portion thereof (e.g., plasma step of a plasma recipe) has achieved a first predetermined result (e.g., the etching away of a certain layer from a multi-layer structure such as a wafer). In this regard, a first plasma process is run in the processing chamber. Optical emissions of the plasma are obtained at a plurality of times during this first plasma process. These optical emissions include at least wavelengths from about 250 nanometers to about 1,000 nanometers (inclusive) which defines a first wavelength range. Optical emissions are preferably obtained at least at every 1 nanometer throughout this first wavelength range. These optical emissions are evaluated or analyzed and at least one endpoint indicator is selected based upon this analysis.

Various features may be utilized by the ninth aspect of the present invention, and these features may be used alone in relation to this ninth aspect as well as in any combination. For instance, the subject analysis may include generating a plot of intensity versus time for a plurality of individual wavelengths which are within the first wavelength range. Preferably, plots are generated for each wavelength which is available based upon the data collecting resolution of the relevant “collecting” structure (e.g., spectrometer(s)). These plots are analyzed after the conclusion of the running of the plasma process, preferably in view of information as to about what time the first endpoint should have occurred (e.g., calculated based upon knowledge of process conditions and thickness of layer to be etched away). Any wavelength having a plot with a distinctive change in intensity around that time where the first endpoint should have occurred may be identified as a possible endpoint indicator candidate.

Further features of the subject ninth aspect relate to the above-noted plots. Initially, the use of the above-described methodology requires no knowledge of the chemistry involved in the subject plasma process. Instead, data is taken over a large wavelength range and at a data collecting resolution which should include at least one indicator of the first endpoint (e.g., at least one specific wavelength which undergoes a change which corresponds with the occurrence of the first endpoint). A pattern of the plot(s)of the individual wavelength(s) which is selected as being a possible candidate for an endpoint indicator of the first endpoint may in turn be used as the endpoint indicator. Moreover, the subject plot may be defined by an equation or function up to that time at which the first endpoint occurs (e.g., a linear function, a first order polynomial, a second order polynomial). When the current plasma recipe no longer “fits” this function, endpoint may be called. First and second derivatives of this function may provide for a more expeditious determination of endpoint and are contemplated by this ninth aspect as well.

Multiple executions of the same plasma process may be required to increase the confidence level associated with the endpoint indicators which are selected as being indicative of the first endpoint. When the above-noted plots are used, a comparison of the plots between two or more runs may identify a pattern which stays the same, but which undergoes some type of change. This change may be a temporal shift, a shift in the intensity associated with the pattern, a uniform enlargement of the pattern, a uniform reduction in the pattern, or any combination thereof. Patterns which undergo this type of change are an indicator that the corresponding wavelength is in fact indicative of the first endpoint. One “controlled” way of inducing such a shift is to process two or more products having different thicknesses. There should be a temporal shift if the particular wavelength is in fact indicative of the first endpoint in this case. That is, the corresponding plot should have a change which shifts in accordance with the change in thickness.

The analysis used to select at least one indicator of the first endpoint may also include examining the optical emissions to identify the existence of intensity peaks, and determining if any of these intensity peaks at least substantially disappear at about a time where the first endpoint should occur. Any such wavelength associated with these types of intensity peaks may be an indicator of the first endpoint. Similarly, the analysis used to select at least one indicator of the first endpoint may include examining the optical emissions to determine if any intensity peaks develop at about a time when the first endpoint should occur. Any such wavelength associated with these types of intensity peaks may also be an indicator of the first endpoint. In addition, the analysis used to select at least one indicator of the first endpoint may include examining the optical emissions to determine if any intensity peak reaches a steady state at about a time when the first endpoint should occur. Any such wavelength associated with these types of intensity peaks may also be an indicator of the first endpoint. Finally, the analysis used to select at least one indicator of the first endpoint may include examining the optical emissions to determine if any intensity peak, which has been at a steady state, undergoes change at about a time when the first endpoint should have occurred. Any such wavelength associated with these types of intensity peaks may also be an indicator of the first endpoint. Any combination of the foregoing may be used to select an endpoint indicator.

A tenth aspect of the present invention relates to monitoring at least two aspects of a plasma process, one of which may be the “health” of the plasma process and another of which may be at least one endpoint associated with the plasma process. This tenth aspect is applicable to any plasma process, including plasma recipes which are run on product (e.g., production wafers, qualification wafers) in a processing chamber, plasma cleanings (e.g., with or without a wet clean), and conditioning wafer operations.

By further way of introduction, substantially the entirety of the plasma process may be evaluated in relation to its “health”, except possibly the initial portion of the plasma process where the plasma is typically unstable. In contrast, the evaluation of the plasma process in relation to identifying an endpoint need not be initiated until closer to the time at which the subject endpoint should be reached. Moreover, the frequency at which the plasma health is evaluated need not be the same as the frequency at which the evaluation is undertaken to identify the subject endpoint. For instance, the plasma health may be assessed less frequently than the evaluation relating to identifying the subject endpoint.

In a first embodiment of the above-noted tenth aspect, a plasma process is conducted within the processing chamber and at least a first endpoint is associated with this plasma process. The plasma process is monitored to identify the occurrence of the first endpoint. Any endpoint detection technique may be used for this first embodiment of the tenth aspect, including those addressed below in relation to the eleventh through the thirteenth aspects of the present invention. The “condition” of the plasma is also evaluated during, and more preferably throughout the entirety of, the plasma process (again excluding possibly the initial portion when the plasma is typically unstable). One way of defining the “condition” associated with this first embodiment of the tenth aspect is equating the same with the cumulative result of all parameters having an effect on the plasma in the processing chamber. This may be done by evaluating optical emissions of the plasma in the chamber which includes at least wavelengths from about 250 nanometers to about 1,000 nanometers which defines a first wavelength range, and at least at every 1 nanometer throughout this first wavelength range and at least at a plurality of different times during the subject plasma process. Another way of characterizing the monitoring of the “health” of the current plasma process is to determine if it is proceeding in accordance with at least one plasma process previously conducted within the same processing chamber. Therefore, the features discussed above in relation to the seventh aspect of the present invention may be utilized in this tenth aspect of the present invention as well.

A second embodiment of this tenth aspect involves generating a plasma in the processing chamber and running a first plasma step in the chamber. Associated with this first plasma step is a first endpoint which is when the first plasma step has produced a first predetermined result. At least one characteristic of the plasma in the chamber is evaluated during the first plasma step using a first time resolution. Although typically equal increments will be utilized in this evaluation, such is not required by this second embodiment of the tenth aspect. An evaluation is also undertaken to identify an occurrence of the first endpoint using a second time resolution which is different than the first. The above-noted “at least one characteristic” may be the condition of the plasma during the subject plasma process, although it need not be the case.

An eleventh aspect of the present invention generally relates to monitoring a plasma process to identify an occurrence of a first endpoint associated with the plasma process. More specifically, at least two different techniques are used to evaluate the current plasma process to identify the first endpoint in this eleventh aspect. Endpoint may be called when only one of these techniques identifies the occurrence of the first endpoint, or may be called after each of these techniques identifies the occurrence of the first endpoint. This eleventh aspect of the present invention is applicable to any plasma process having at least one endpoint associated therewith (e.g., plasma recipes which are run on product in a processing chamber, plasma cleanings, and conditioning wafer operations).

One of the techniques which may be used in the subject eleventh aspect involves a comparison of the current optical emissions of the plasma in the chamber with optical emissions of the plasma in the chamber from a previous time in the same process, preferably the immediately preceding time at which optical emissions were obtained. In one embodiment, these optical emissions include at least wavelengths from about 250 nanometers to about 1,000 nanometers at least at about every 1 nanometer. When these optical emissions are substantially a “match” (e.g., when the differential of the pattern of current optical emissions and the pattern of previous-in-time optical emissions is substantially free of peaks), particularly after an initial portion of the plasma has been completed, endpoint may be deemed to have been reached. Stated another way, when there is no longer any substantial change in the optical emissions, endpoint may be deemed to have been reached.

Another technique for identifying endpoint which may be used in the subject eleventh aspect involves a comparison of the current optical emissions of the plasma in the chamber with a standard. This “standard” may be optical emissions of the plasma in the chamber from a previous execution of this same plasma process in the same processing chamber at a time when endpoint was at least deemed to have been reached. Moreover, this standard may be stored on a computer-readable storage medium. In one embodiment, these optical emissions include at least wavelengths from about 250 nanometers to about 1,000 nanometers at least at about every 1 nanometer. When these optical emissions are substantially a “match” (e.g., when the differential of the pattern of current optical emissions and the pattern of previous-in-time optical emissions is substantially free of peaks), particularly after an initial portion of the plasma has been completed, endpoint may be deemed to have been reached.

Yet another technique which may be used in the subject eleventh aspect of the present invention includes determining if there is at least a first change in the impedance of the processing chamber which is reflected in the optical emissions of the plasma in the processing chamber. A “modal” change in the plasma may be indicative of a change in impedance which in turn is indicative of endpoint. This “modal” change may be a rather sudden and significant increase or decrease in the intensity of the entirety of the plasma or of a particular wavelength(s).

Another technique which may be used to identify endpoint in relation to the subject eleventh aspect includes evaluating at least one individual wavelength of light forming the plasma of the subject plasma process. This one wavelength of light may be evaluated to determine when a plot of intensity versus time deviates by more than a predetermined amount from a predetermined equation (e.g., when there is no longer a “fit” between the current data and the subject equation). Therefore, the features discussed above in relation to the ninth aspect of the present invention are also relevant to this portion of the eleventh aspect as well. Moreover, any one or more individual wavelengths of light may be evaluated to determine when the change in slope over time of the wavelength(s) changes by more than a predetermined amount. Second order derivatives may be used as well.

A twelfth aspect of the present invention is directed toward a technique for identifying the occurrence of a first endpoint associated with a plasma process (e.g., plasma recipe, plasma clean, conditioning wafer operation) or a discrete/discernible portion thereof (e.g., a plasma step of a multiple step recipe or process). Optical emissions of the plasma in the chamber from the process are obtained. These optical emissions include at least wavelengths from about 250 nanometers to about 1,000 nanometers which defines a first wavelength range. The data resolution which is used in collecting the optical emissions is no more than about 1 nanometer. This means that optical emissions are obtained at least at every 1 nanometer throughout the first wavelength range.

Identification of the first endpoint involves a comparison of the most current optical emissions of the plasma in the chamber with a first output. This first output may be optical emissions of the plasma in the chamber from a previous time in the same plasma process, preferably the immediately preceding time at which the optical emissions were obtained in relation to the now current optical emissions. This first output may also be optical emissions from the same type of plasma process which was previously conducted in the same processing at a time when endpoint should have occurred. In this case the first output could be stored on a computer-readable storage medium. When the above-noted comparison indicates that the current optical emissions and the first output are at least substantially a “match”, particularly after an initial portion of the plasma has been completed, the first endpoint is deemed to have been reached. Confidence in the calling of the first endpoint by the above-noted technique may be enhanced by using a second technique and not calling the first endpoint until both of the techniques have “seen” the first endpoint. Any of the techniques discussed above in relation to the eleventh aspect of the invention may be utilized in this twelfth aspect for this purpose.

A thirteenth aspect of the present invention relates to identifying the occurrence of multiple endpoints in a single plasma process. Many plasma recipes will include a number of different plasma steps. Each of these plasma steps typically has an identifiable endpoint associated therewith. Therefore, the eleventh aspect of the present invention allows for the identification of at least two of these endpoints and including each endpoint associated with the subject plasma process. Each of the techniques identified in eleventh aspect discussed above may be utilized in this thirteenth aspect.

When to clean a processing chamber is the subject of the fourteenth aspect of the present invention. Product is loaded in the processing chamber. The processing chamber is sealed and a first plasma process is thereafter run on the product. Data relating to the plasma process is obtained. From this data a determination is made regarding the condition of the chamber. Specifically, a determination is made as to whether the interior of the chamber is sufficiently “dirty” from plasma processes previously run in the chamber to warrant cleaning the chamber. Personnel may be notified, actions may be initiated, or both, if this “dirty chamber” condition is identified. Appropriate actions include terminating the current plasma process, issuing an alert, suspending execution of any further plasma processes in the chamber until it is appropriately cleaned, or any combination thereof.

Various features may be utilized by the fourteenth aspect of the present invention, and these features may be used alone in relation to this fourteenth aspect as well as in any combination. The data which is obtained on the current plasma process may be optical emissions of the plasma in the chamber. Wavelengths obtained may include at least from about 250 nanometers to about 1,000 nanometers which defines a first wavelength range. Data may be obtained at least at every 1 nanometer throughout the noted first wavelength range.

Numerous techniques may be implemented to determine if the processing chamber should be cleaned using the data obtained on the plasma in the chamber. Description of these techniques will be in relation to optical emissions data. The current optical emissions (from the current time in the process) may be compared with a standard which is stored on a computer-readable storage medium. This standard may be optical emissions of plasma from the same chamber, but from a plasma process previously run in the chamber where the chamber was determined or assumed to be in need of cleaning. When the current optical emissions are at least substantially a match with this standard, the chamber will be deemed to be in need of cleaning. “Matches” may be based upon any pattern recognition technique. Determining the differential between the current optical emissions and the standard may also be used for this purpose.

Another way of determining when to clean the chamber using data obtained on the plasma involves endpoint detection. Each of the endpoint detection techniques discussed above in relation to the tenth through the thirteenth aspects of the present invention use data relating to the plasma in the chamber to identify endpoint. When any step of a multiple step plasma process takes longer than a pre-set maximum time limit, the chamber may be deemed to be in need of cleaning. If the total time to complete an entire plasma process takes longer than a pre-set maximum time limit, the chamber may also be deemed to be in need of cleaning. Endpoint detection techniques may be used in each of these cases. In the case of a multiple step plasma process, the endpoint may be identified for each of the steps of a multiple step process, or simply the endpoint of the last step in the subject plasma process may be identified, to determine the total time spent on the process.

Plasma cleaning operations are embodied within a fifteenth aspect of the present invention. A plasma clean removes materials from the interior of the processing chamber by having plasma exist within an “empty” chamber. No product (e.g., wafers) is contained within the chamber during a plasma clean.

Optical emissions of the plasma in the “empty” chamber are obtained at a plurality of times during the process in a first embodiment of this fifteenth aspect. A pattern of at least a portion of the optical emissions is compared with a first standard pattern during at least a portion of the process (e.g., a plot of intensity versus time). This first standard pattern may be recorded on a computer-readable storage medium. Furthermore, this first standard pattern may be from optical emissions of plasma in a plasma process previously conducted within the same processing chamber at a time when the plasma clean had reached its endpoint. When the pattern of optical emissions from at least one time during the process and the first standard pattern are within a predetermined amount of each other, the plasma process is terminated. “Predetermined amount” contemplates using pattern recognition techniques, as well as taking a differential and noting when this differential is at least substantially free from any substantial intensity peaks.

Various features may be utilized by the first embodiment of the fifteenth aspect of the present invention, and these features may be used alone in relation to this fifteenth aspect as well as in any combination. Wavelengths including at least those from about 250 nanometers to about 1,000 nanometers (first wavelength range), at least at every 1 nanometer throughout the first wavelength range, may be obtained and utilized for the comparison with the first standard pattern. All of these optical emissions may be utilized, or only a portion thereof. That is, the first embodiment of this fifteenth aspect includes comparing the pattern of a specific wavelength(s) within the optical emissions of the plasma with the first standard pattern which will include the corresponding wavelength(s). Moreover, the first embodiment also includes comparing the pattern of the entirety of the optical emissions obtained on the current plasma process with the first standard pattern.

Difficulties may be encountered when using a particular wavelength for the first standard pattern. One such difficulty may be finding this particular wavelength in the optical emissions of the current plasma process due, for instance, to a wavelength shift. Additional features may be utilized in the first embodiment of the fifteenth aspect to address this type of a situation. In this regard, the first standard pattern may be part of a first standard optical emissions segment which includes a plurality of wavelengths. The intensity peak associated with the subject wavelength of the first standard pattern may be identified in relation to its intensity (e.g., it is the “largest” intensity peak around a certain wavelength), one or more other intensity peaks (e.g., the subject wavelength is the “middle” peak in a certain wavelength region), or both. Noting these characteristics of the wavelength for the first standard pattern in the first standard optical emissions segment may then be used to identify the subject wavelength in the current optical emissions segment.

The first embodiment of the subject fifteenth aspect may also be terminated if the subject plasma process has reached a predetermined maximum time limit before the current pattern and first standard pattern are within a predetermined amount of each other. Typically this will mean that the current plasma process was ineffective in addressing the interior of the processing chamber. In this type of a case, a wet clean of the chamber may be initiated. Thereafter, another plasma cleaning operation may be initiated to address the residuals of the wet clean.

Monitoring the time rate of change of optical emissions of the plasma in the chamber is the subject of a second embodiment of the fifteenth aspect relating to plasma cleans. In this regard, the differential between the optical emissions at a current time in the process and the optical emissions from a previous time in the same plasma process (preferably an immediately preceding time) is determined. When this differential is no more than a first amount, the current plasma process is terminated. Therefore, this second embodiment equates the time at which the plasma clean should be terminated with a situation where the current plasma process is no longer changing the condition of the interior of the processing chamber at a desired rate. All or a portion of those “additional” features addressed above in relation to the first embodiment may be implemented in this second embodiment as well.

Conditioning wafer operations are addressed in a sixteenth aspect of the present invention. At least one conditioning wafer is loaded in a processing chamber and a plasma process is run thereon. Typically the plasma process will etch a pattern on the conditioning wafer which is something other than an integrated circuit or a pattern which would not be associated with a semiconductor device. Plasma processing of the conditioning wafer is monitored through obtaining optical emissions of the plasma in the chamber. A number of conditioning wafers are processed in this manner until the conditioning wafer operation is terminated based upon the results of the monitoring of one of the plasma processes conducted on a conditioning wafer. Thereafter, a production wafer operation is initiated whereby at least one production wafer is loaded in the chamber and a plasma recipe (e.g., one or more plasma steps) is run thereon. These production wafers are removed from the chamber and at least one semiconductor device is formed therefrom. Further processing of the production wafer may be required before the actual semiconductor device is available. Therefore, this distinguishes a production wafer from a conditioning wafer since no semiconductor devices are formed from conditioning wafers. Instead, conditioning wafers are either typically scrapped or refurbished for further use as a conditioning wafer.

Various features may be utilized by the sixteenth aspect of the present invention, and these features may be used alone in relation to this sixteenth aspect as well as in any combination. The health of the conditioning wafer operation, the health of the production wafer operation, or both may be evaluated. Any of the techniques discussed above in relation to the seventh and tenth aspects of the invention thereby may be implemented in this sixteenth aspect as well. Optical emissions obtained on the conditioning wafer operation, the production wafer operation, or both may include at least wavelengths from about 250 nanometers to about 1,000 nanometers, and such optical emissions may be obtained at least at every 1 nanometer throughout this range. All or a portion of this data may be utilized in a comparison upon which termination of the conditioning wafer operation is based.

Endpoint detection techniques may be used to terminate the conditioning wafer operation. Therefore, any of the endpoint detection techniques discussed above in relation to the ninth through the thirteenth aspects may be implemented in this sixteenth aspect as well. Termination of the conditioning wafer operation may also be based upon when consecutive runnings of the plasma process on conditioning wafers are within a certain amount of each other as determined through the data obtained on the process. That is, the termination of the conditioning wafer operation may be equated with the conditioning wafer operation having reached a steady state (e.g., the processing of one conditioning wafer looks at least effectively the same as the processing of the next conditioning wafer) determined in accordance with an evaluation of optical emissions data. Termination of the conditioning wafer operation may also be based solely on the data obtained on the conditioning wafer operation. That is, no wafer need be analyzed before the production wafer operation is initiated. One or more of a plasma cleaning operation, a wet cleaning operation, or a replacement of consumables may also be initiated before the initiation of the conditioning wafer operation as well.

Management of the distribution of wafers to a wafer production system including at least two processing chambers is addressed by a seventeenth aspect of the present invention. In a first embodiment, at least two chambers are involved in the plasma processing of wafers disposed therein. Each plasma process conducted within these chambers is monitored in at least some respect. Wafers will continue to be sequentially processed in these chambers unless the monitoring of the current plasma process on the wafer(s) in one of these chambers detects the existence of one or more conditions. These conditions include the existence of a “dirty chamber”, a known error condition, an unknown condition, or a combination thereof as these terns have been used in relation to the sixth and fourteenth aspects discussed above. The distribution of wafers to this particular chamber may be suspended immediately after this type of condition is identified, or suspension may be delayed until a certain number of these types of conditions are encountered in multiple plasma processes. That is, a given chamber may not be taken “off line” until this same condition (or another of the conditions) have been identified in multiple runs.

When the suspension of processing of wafers in a given chamber is based upon the identification of a “dirty chamber” condition, the chamber may be cleaned in some manner. Plasma cleans, wet cleans, replacement of consumables, or any combination thereof are contemplated as an appropriate “cleaning” in the context of this first embodiment of the seventeenth aspect. Once the chamber has been cleaned, the distribution of wafers for running plasma processes thereon may be reinitiated. Encountering a known error during the plasma processing of a wafer(s) in one of the chambers may result in the modification of one or more process control parameters to address this error. Finally, when an unknown condition is encountered, the first embodiment contemplates analyzing the plasma process after termination thereof in an attempt to identify the corresponding cause.

A second embodiment of the seventeenth aspect relates to the running of plasma processes on product in at least three chambers. The wafers are distributed to these chambers using a first sequence. Modification of this sequence is initiated if the monitoring of the plasma process in one of the chambers identifies the existence of a certain condition. Any of those identified above in relation to the first embodiment would be applicable to this second embodiment as well. In this regard, the corresponding features from the first embodiment may be implemented in this second embodiment as well.

Finally, a third embodiment of this seventeenth aspect involves the distribution of wafers to at least two processing chambers for the running of a plasma process thereon. The time required to complete each plasma process is monitored. The distribution sequence which is utilized is based upon this monitoring of time. For instance, the distribution sequence may involve maximizing the use of the “fastest” processing chamber.

The present invention will now be described in relation to the accompanying drawings which assist in illustrating its various pertinent features. One application of the present invention is for processes which utilize plasma to provide at least one function or to achieve at least one predetermined result, and the present invention will hereafter be described in this context. More specifically, the present invention will be described in relation to the running of plasma processes on wafers or the like from which semiconductor devices are formed (e.g., etching where the “predetermined result” may be the removal of one or more layers, chemical vapor deposition where the predetermined result may be the buildup of one or more films, sputtering where the predetermined result may be the addition or removal of material).

Wafer Production System2—FIG.1

Awafer production system2 is illustrated in FIG.1 and is generally for executing one or more plasma-based processes (single or multiple step) onwafers18. Semiconductor devices may be formed fromwafers18 processed by thesystem2, including integrated circuit chips. Thesystem2 generally includes awafer cassette6 which stores a plurality ofwafers18 and allows thesewafers18 to be readily transported to and from thesystem2. Onewafer cassette6 is disposed in each of the two load lock chambers28 of thewafer production system2. Awafer handling assembly44 is advanced into the respective load lock chamber28, removes at least one of thewafers18 from thewafer cassette6, and transfers the wafer(s)18 to one of the plurality ofprocessing chambers36 of the wafer production system2 (fourchambers36a-dbeing illustrated). Other arrangements may be utilized for purposes of the present invention.

Control of thewafer handling assembly44, as well as various other components of thewafer production system2, is provided by a main control unit58 (hereafter “MCU58 ”). In one embodiment, theMCU58 is a computer having at least one computer-readable storage medium and at least one processor, such as a desktop PC or a main-frame having satellite terminals. Appropriately integrating theMCU58 with one or more other components of thewafer production system2 allows theMCU58 to be the main controller for thechamber36. Integration may include operatively interfacing theMCU58 with these various components and including thewafer handling assembly44. Other hardware may also be operatively interconnected with theMCU58, such as adisplay59 for providing visual-based information to operations personnel (e.g., a CRT or computer monitor), as well as a data entry device60 (e.g., mouse, light pen, keyboard) for allowing operations personnel to enter information used by or relating to thewafer production system2.

Plasma within thechambers36 processes the enclosed wafer(s)18 in some manner (e.g., etching to remove a predetermined layer of material). Atransparent window38 is provided on eachchamber36 to allow optical emissions data to be obtained on the plasma recipe being run on the wafer(s)18 in therespective chamber36. Once the plasma process is completed, thewafer handling assembly44 removes the wafer(s)18 from therespective processing chamber36 and transfers the wafer(s)18 back to one of thewafer cassettes6 in the associated load lock chamber28. When all of thewafers18 within one of thecassettes6 have been plasma processed, thewafer cassette6 is removed from the load lock chamber28 and replaced with anothercassette6 withnew wafers18 to be plasma processed. This may be done manually by operations personnel or in an automated manner by robots or the like.

Wafer Cassette6—FIG.2

More details regarding the embodiment of thewafer cassette6 which is incorporated in thewafer production system2 of FIG. 1 are presented in FIG.2. Thewafer cassette6 includes aframe10 defined by a pair of laterally spaced sidewalls22 which are interconnected by aback panel26, as well as a pair ofend panels8. The front of theframe10 is substantially open such that thewafer handling assembly44 may be advanced within and retracted from thewafer cassette6 to removewafers18 from and providewafers18 to the associatedwafer cassette6. A plurality of longitudinally spaced and laterally disposed partitions16 (e.g., eachpartition16 being disposed at least generally perpendicular to the longitudinal axis of the cassette6) are provided within theframe10 for purposes of maintaining separation ofadjacent wafers18. Each pair ofadjacent partitions16 defines apocket14 in which asingle wafer18 may be placed. Loading ofwafers18 within thewafer cassette6 which are to be plasma processed may be accomplished by disposing the one of theend panels8 of thecassette6 on an appropriate supporting surface and manually loadingwafers18 into thecassette6, with only onewafer18 being disposed in any of thepockets14. Once thewafer cassette6 is loaded withwafers18 to the desired degree, thecassette6 may be transported to the appropriate load lock chamber28. Thewafer cassette6 is then disposed on one of itsends8 in a position such that its substantially open front faces and is accessible by thewafer handling assembly44. Other configurations forwafer cassettes6 may be utilized by thewafer production system2, and automation may be implemented in any one or more of the loading ofwafers18 into thecassette6 and the transport of thecassette6 to and from the load lock chambers28 of thewafer production system2.

Wafer Handling Assembly44—FIGS.3A-3B

Additional details regarding thewafer handling assembly44 which is incorporated in thewafer production system2 of FIG. 1 are illustrated in FIGS. 3A-B. Other types of wafer handling assemblies may be utilized by thewafer production system2, such as the types disclosed in U.S. Pat. No. 5,280,983 to Maydan et al., issued Jan. 25, 1994, and entitled “SEMICONDUCTOR PROCESSING SYSTEM WITH ROBOTIC AUTOLOADER AND LOCK” and U.S. Pat. No. 5,656,902 to Lowrance, issued Aug. 12, 1997, and entitled “TWO-AXIS MAGNETICALLY COUPLED ROBOT”, both of which patents are incorporated by reference in their entirety herein. Thewafer handling assembly44 of FIGS. 3A-B generally includes arobotic wafer handler48 which is disposed within acentral chamber70 of thewafer production system2. Load lock chambers28 andprocessing chambers36 are thereby disposed about thewafer handling assembly44. Movement of therobotic wafer handler48 is realized through a waferhandler control motor62 which is operatively interfaced with thewafer handler48, and which in turn is operatively interfaced with and controlled by the MCU58 (FIG.1). Thewafer handler48 includes apivot50 such that thewafer handler48 may be pivoted or rotated about thepivot50 to position thewafer handling assembly44 to appropriately interface with one of the load lock chambers28 orprocessing chambers36.Wafers18 are removed from and provided to the respective load lock chamber28 orprocessing chamber36 by awafer blade66 which interfaces with one of thewafers18 when disposed in one of thepockets14 of thewafer cassette6. A vacuum chuck or the like may be incorporated on thewafer blade66 to retain thewafer18 on the blade66 (not shown). Thewafer blade66 is interconnected with anarm assembly54 which extends and retracts via pivotal-like action to axially advance and retract thewafer blade66 to the appropriate position.

Processing Chamber72—FIGS.4 and5

One embodiment of a processing chamber which may be incorporated in the wafer production system of FIG. 1 as one of thechambers36 is presented in more detail in FIG.4. Other types/configurations of processing chambers may be utilized by thewafer production system2 for purposes of the present invention as well, including those disclosed in U.S. Pat. No. 5,614,055 to Fairbairn et al., issued Mar. 25, 1997, and entitled “HIGH DENSITY PLASMA CVD AND ETCHING REACTOR”, and U.S. Pat. No. 5,641,375 to Nitescu et al., issued Jun. 24, 1997, and entitled “PLASMA ETCHING REACTOR WITH SURFACE PROTECTION MEANS AGAINST EROSION OF WALLS”, both patents of which are incorporated by reference in their entirety herein.

Theprocessing chamber74 of FIG. 4 is specifically adapted for performing a plasma etching operation on a wafer(s)18 when disposed therein (i.e., to remove one or more layers from the wafer(s)18 disposed in the chamber36). Theprocessing chamber74 includes chamber sidewalls78 which are disposed about a central,longitudinal axis76 of thechamber74. Access to theprocessing chamber74 may be provided by achamber cover82 which is interconnected with the chamber sidewalls78 in such a manner that at least in certain instances, at least a portion of thechamber cover82 may be moved away from thechamber sidewalls78. In the illustrated embodiment, thechamber cover82 is removed only to gain access to the interior of theprocessing chamber74 for maintenance, cleaning, or both. Awindow port124 extends through a portion of thechamber sidewall78 and is aligned with atransparent window112. Thewindow112 includes aninner surface116 and anouter surface120, and provides a way for the plasma to be viewed exteriorly of theprocessing chamber74 and further to provide a mechanism for obtaining optical emissions data on the plasma recipe being run on the wafer(s)18 within thechamber74.

Protection of the chamber sidewalls78 and the chamber cover82 from the effects of plasma processes being conducted within thechamber74 is provided by abell jar90 and abell roof86 which are each formed from transparent, dielectric materials (e.g., quartz, sapphire). Thebell jar90 is spaced radially inward (e.g., in the direction of the central,longitudinal axis76 of the chamber74) from the inner surface of thechamber sidewalls78. Thebell roof86 is disposed above thebell jar90 and is axially movable in a direction which is at least substantially parallel with the central,longitudinal axis76 of thechamber74 through interconnection with anelevator98. Movement of theelevator98 may be desirable for one or more purposes. For instance, this movement may be used to change the spacing between a showerhead94 and awafer pedestal106/wafer platform102 which in one embodiment are the electrodes or “plasma generator” for thechamber74.

Thewafer pedestal106 is disposed radially inwardly of thebell jar90 in spaced relation therewith, and thewafer platform104 is disposed on top of thewafer pedestal106. In one embodiment, both thewafer pedestal104 andwafer platform106 are formed from silicon-based materials since thewafers18 are also commonly formed from silicon-based materials. Thewafer18 is introduced into theprocessing chamber74 through awafer access80 which extends through thechamber sidewall78, and is disposed in a flat orientation on the upper surface of thewafer platform104. Various mechanisms may be used to retain thewafer18 on thewafer platform104 during the running of the plasma process on thewafer18 in thechamber74, such as by drawing a vacuum through avacuum port108 which is formed on thewafer platform104 or by using electrostatic charges (not shown). Transport of thewafer18 into theprocessing chamber74 is again provided by thewafer blade66 of the wafer handling assembly44 (FIGS.1 and3A-B). After thewafer blade66 is retracted from theprocessing chamber74, a vacuum is generated within theprocessing chamber74 before the plasma process is initiated.

The showerhead94 is interconnected with theelevator98 such that it is axially movable therewith, and in one embodiment is also formed from a silicon-based material for the above-noted reasoning. The showerhead94 includes one or more apertures (not shown) for the purpose of dispersing feed gases within the vacuum chamber84 in a manner to define a desired gas flow pattern for the plasma. Gases are provided to the showerhead94 through agas inlet port100 formed in thequartz bell roof86. With appropriate gases being contained within theprocessing chamber74 and under other appropriate conditions (e.g., pressure, temperature, flow rate), an appropriate voltage may be applied to one or more of thewafer pedestal106 and the showerhead94 to create the plasma within thechamber74 above thewafer platform104. Thewafer pedestal106 andwafer platform104, as well as the showerhead94, thereby also function as electrodes in the illustrated embodiment as noted. The electrical field generated by these electrodes also functions to effectively confine the plasma to the space between the electrodes.

FIG. 5 illustrates one embodiment of agas delivery system150 which may be used to provide gases to theprocessing chamber74 of FIG. 4 for a given plasma process operation. Other systems may be utilized as well. Thegas delivery system150 includes a plurality of storage tanks154, each of which is fluidly interconnected with theprocessing chamber74 either directly or indirectly. Storage tanks154b-dare available for containing one or more types of feed gases which will define the gas composition of the plasma within the vacuum chamber84. Each storage tank154b-dis fluidly interconnected with amixer166 by gas lines158b-dwhere the feed gases may be appropriately mixed prior to being provided to theprocessing chamber74 through the showerhead94 via thegas line158e.Mixing of the feed gases could also occur in a manifold (not shown) into which each of the feed gases would separately flow and which could be contained within or be part of theprocessing chamber74. The manifold would then interface with thegas inlet port100, and this type of manifold may also be used in combination with themixer166. In some cases the composition of the feed gases provided to theprocessing chamber74 to define the plasma may be difficult to ignite. This situation is remedied by including an appropriate gas in thestorage tank154a.A gas composition which is more readily ignitable than the composition of the feed gases is contained within thestorage tank154a.Ignition of the plasma would then be affected by directing a flow of the ignition gas from thestorage tank154ainto theprocessing chamber74, along with a flow of the desired feed gases from the storage tanks154b-d,and using the ignition of the ignition gas to then ignite the feed gases to generate the plasma.

Plasma Monitoring Assembly174—FIGS.6 and7

The above-described components are obviously important to the overall function of thewafer production system2. However, the present invention is more specifically directed to the monitoring or evaluation of the plasma itself. Therefore, the following components may be incorporated in any type of plasma-based system, including the foregoing.

One embodiment of an assembly for monitoring/evaluating plasma processes and which may be incorporated in thewafer production system2 of FIG. 1 is illustrated in FIG.6. Theplasma monitoring assembly174 operatively interfaces with thewindow38 of theprocessing chamber36 by receiving optical emissions of the plasma which pass out of theprocessing chamber36 through thewindow38. These optical emissions are “collected” by an appropriatefiber optic cable178 which is positioned at or near theouter surface42 of thewindow38. Fixtures which illustrate ways of maintaining a fiber optic cable and a window of a processing chamber in a fixed positional relationship are presented in FIGS. 36 and 39. Optical emissions of the plasma within theprocessing chamber36 during processing of awafer18 enter thefiber optic cable178 and are directed to aspectrometer assembly182. Both scanning-type and solid state spectrometers may be used as thespectrometer assembly182. Theassembly182 may also include one or more appropriately interconnected spectrometers, each of which obtains optical emissions data from a different region. Thespectrometer assembly182 separates these optical emissions into a plurality of individual wavelengths and provides these separate optical components to anarray186 of charge coupled devices (hereafter “CCD array186”) for conversion to a corresponding electrical signal.

A computer-readable signal is provided by theCCD array186 to a plasma monitor control unit128 (hereafter “PMCU128”) which is the primary control mechanism of theplasma monitoring assembly174. In one embodiment, thePMCU128 is a computer which may be configured to include, but not limited to, at least one motherboard, at least one analog-to-digital conversion board, at least one central processing unit (CPU) for each motherboard, and one or more types of computer-readable storage mediums such as at least one floppy disk drive, at least one hard disk drive, and at least one CD ROM drive. Other hardware may be operatively interconnected with thePMCU128, such as adisplay130 for providing visual/audio-based information to operations personnel (e.g., a CRT, LCD, or computer monitor), as well as one or more data entry devices132 (e.g., mouse, light pen, keyboard) for allowing personnel to enter information used by or relating to theplasma monitoring assembly174. OnePMCU128 may be provided for eachchamber36, or thePMCU128 may be configured to servicemultiple chambers36. ThePMCU128 is also operatively interfaced or interconnected with theMCU58 of thewafer production system2 such that the PMCU andMCU58 may communicate with each other.

ThePMCU128 includes aplasma monitoring module200 and each of its sub-modules may be stored on a computer-readable storage medium associated with the PMCU128 (e.g., on a portable computer diskette(s), on a hard drive, on a CD(s)). Theplasma monitoring module200 and these sub-modules are illustrated in FIG.7. One sub-module is astartup module202 which provides a way of accessing other sub-modules through a currentplasma process module250. The currentplasma process module250 of theplasma monitoring module200 facilitates the monitoring or evaluation of the various types of plasma processes which may be conducted within thechamber36 through the evaluation of optical emissions data of the plasma in thechamber36. In the case of the FIG. 6 embodiment, optical emissions data are collected and delivered by thefiber optic cable178 to thespectrometer assembly182 which divides the light up into its individual optical components. Data representative of these optical emission components are then made available to the currentplasma process module250 through theCCD array186 as described above.

Evaluation or monitoring of the current plasma process through the currentplasma process module250 is facilitated by collecting optical emissions from the plasma preferably to include at least wavelengths from within the UV range to within the near infrared range, and thereby including the visible light spectrum. In one embodiment, optical emissions of the plasma in theprocessing chamber36 which are obtained and available for evaluation (e.g., by the currentplasma process module250, manually by the appropriate personnel) include at least those wavelengths from about 250 nanometers to about 1,000 nanometers (inclusive), and more preferably at least those wavelengths from about 150 nanometers to about 1,200 nanometers (inclusive). Hereafter the above-noted desired range or bandwidth of optical emissions data which are obtained/collected of the plasma in thechamber36, and which includes each of the above-noted ranges or bandwidths, will be referred to as the “Preferred Optical Bandwidth.”

Optical or wavelength resolutions within and throughout the Preferred Optical Bandwidth are preferably no more than about 1 nanometer, and even more preferably no more than about 0.5 nanometers (presently contemplating a wavelength resolution of 0.4). The term “wavelength resolution” in this context means the amount of separation between adjacent wavelengths in the subject optical emissions data which is collected. Therefore, if the wavelength resolution being used to collect optical emissions data from the plasma in thechamber36 is 1 nanometer, no more than a 1 nanometer spacing will exist between any two data points within and throughout the Preferred Optical Bandwidth. Although equal spacings will typically be utilized in relation to the wavelength resolution within and throughout the Preferred Optical Bandwidth, this need not be the case such that “wavelength resolution” encompasses equal spacings, unequal spacings, and combinations thereof. Hereafter, the above-noted magnitude for the optical or wavelength resolution will be referred to as the “Preferred Data Resolution.”

Another factor relating to the effectiveness of the currentplasma process module250 in relation to the amount of optical emissions data of the plasma in thechamber36 is the times at which this data is taken during the subject plasma process. Optical emissions data of the plasma in thechamber36 are preferably obtained at least every 1 second, and more preferably at least every ⅓ second. Although equal spacings will typically be utilized in relation to the times at which optical emissions data is collected on the plasma in thechamber36, this need not be the case such that equal time spacings, unequal spacings, and combinations thereof may be utilized. Hereafter, the above-noted timing magnitudes for obtaining optical emissions data of the plasma in thechamber36 will be referred to as the “Preferred Data Collection Time Resolution.”

Thespectrometer assembly182 illustrated in FIG. 6 should be capable of meeting the above-noted criteria, and a number of implementations may be utilized. For instance, thespectrometer assembly182 may be of the scanning type in which thespectrometer assembly182 would include structure to scan the spectrum to obtain data encompassing the Preferred Optical Bandwidth using the Preferred Data Resolution and at the Preferred Data Collection Time Resolution (e.g., scan a first optical emissions segment or region of the 250-550 nanometer wavelengths, scan a second segment of the 500-750 nanometer wavelengths, and scan a third segment of the 700-950 nanometer wavelengths, each of which overlaps so that the possibility of losing data is reduced and further to facilitate alignment of spectral segments). Thespectrometer assembly182 may also be a solid state device. Multiple subunits or processing cards may be connected in parallel relation to function similar to the scanning type noted above. That is, each subunit or processing card of the solid state device would then provide information on a specific optical emissions segment or region within the Preferred Optical Bandwidth using the Preferred Data Resolution and at the Preferred Data Collection Time Resolution.

Exemplary Plasma Recipe Spectra—FIG.8

A representative or exemplary spectra in computer-readable form which may be made available to the currentplasma process module200 of FIG. 7 for analysis is presented in FIG.8. Only a portion of the Preferred Optical Bandwidth is represented by thespectra246. However, it serves to illustrate certain principles associated with the present invention since the evaluation of the current plasma process need not be of the entire Preferred Optical Bandwidth in each case. Thespectra246 contains data within the wavelength range of 400 nanometers to about 700 nanometers and at a certain fixed point in time in a plasma process being conducted within theprocessing chamber36 of FIG. 1 (e.g., at a current time t_n). Various characteristics of thespectra246 of FIG. 8 may be used in the analysis undertaken by the currentplasma process module250 of FIG.7. These characteristics include the overall pattern of thespectra246, one or more of the location and intensity of one or more of intensity peaks248 in thespectra246, and one or more of the relative location and relative intensity of one or more of the intensity peaks248.

The currentplasma process module250 of FIG. 7 operatively interfaces with a collection of spectra obtained from one or more plasma processes previously conducted within thesubject processing chamber36. Generally, the currentplasma process module250 receives data from the current plasma process being conducted within theplasma processing chamber36, and in all but one case (the research module1300 ) compares this data or at least a portion thereof with data from one or more plasma processes previously conducted within this very sameplasma processing chamber36 to evaluate or monitor the current plasma process. This comparison provides certain information regarding the current plasma process depending upon which sub-module(s) of the currentplasma process module250 is being utilized. Each of these sub-modules will be discussed in more detail below in relation to the relevant figures. However, understanding how the data is organized for access by the currentplasma process module250 may facilitate a more complete understanding of these sub-modules.

Plasma Spectra Directory284—FIGS.9-12B

One embodiment of how the collection of data on prior plasma processes may be organized for use by the currentplasma process module250 is illustrated in FIG.9. Theplasma spectra directory284 of FIG. 9 includes a number of subdirectories or subsets of categorically similar data, is typically specific to a single processing chamber36 (although thesame directory284 could be used formultiple chambers36 if the data within thedirectory284 was indexed in some way to the specific chamber36), is accessed by one or more of the submodules of the currentplasma process module250, and is preferably stored in a computer-readable medium of or associated with the PMCU (e.g., one or more computer diskettes, hard drive, one or more CDs). Relevant data which is included in each data entry in each of these subdirectories of theplasma spectra directory284 is a spectra (optical emissions data), and in all but one case (spectra of calibration light subdirectory310) is a spectra of the plasma in theprocessing chamber36 at the relevant time and typically within the Preferred Optical Bandwidth and at the Preferred Data Resolution unless otherwise noted herein.

Plasma processes which are to be used as a standard of sorts for evaluating/monitoring plasma processes currently being run in theprocessing chamber36, or to assess the health or condition of the plasma during such plasma processes, are stored in a subdirectory of theplasma spectra directory284 of FIG. 9 which is entitled the “Spectra of “Normal” Plasma Processes” and identified by reference numeral288 (hereafter “normal spectra subdirectory288”). Actual spectral data from one or more plasma processes which have been conducted in theprocessing chamber36 are stored in thenormal spectra subdirectory288 associated with thischamber36. Various categories or genera of plasma processes may be stored in thenormal spectra subdirectory288, such as the above-noted plasma recipes, plasma cleaning operations or plasma cleans (with or without previously conducting a wet clean), and conditioning wafer operations, and one or more executions of plasma processes within each of these categories may also be stored in thesubdirectory288. Each category of plasma process may be organized in a “folder” or the like of thenormal spectra subdirectory288, or may have a code which identifies it as being representative of a particular category/genus of plasma process to group categorically similar processes together (e.g, a folder for “normal” plasma recipes, a separate folder for “normal” plasma cleans which are run without first wet cleaning thechamber36, a separate folder for “normal” plasma cleans which are run after a wet clean of thechamber36, and a separate folder for “normal” conditioning wafer operations).

Spectral data is used by the currentplasma process module250 to determine if subsequent plasma processes conducted in this verysame processing chamber36 are proceeding in accordance with at least one of the plasma processes stored in thenormal spectra subdirectory288. Entries in thenormal spectra subdirectory288 are thereby used as a “model” or “standard” for the evaluation of plasma processes conducted in this verysame processing chamber36 at some future time. How data is actually entered in thenormal spectra subdirectory288 will be discussed in more detail below in relation to thestartup module202 and FIGS. 13-14. Suffice it to say for present purposes that entries in thenormal spectra subdirectory288 are from actual plasma processes conducted in thesubject chamber36. These plasma processes are either confirmed (e.g., by post-plasma processing testing) or assumed (and typically later confirmed) to have proceeded in a desired or predetermined manner, or more specifically without any substantial/significant errors or aberrations. No pre-analysis or knowledge of any plasma process is required to use the currentplasma process module250 and thenormal spectra subdirectory288 to evaluate a plasma process currently being conducted in theprocessing chamber36. Spectral data from a plasma process ABC conducted in a givenchamber36 may be recorded in thenormal spectra subdirectory288 one day simply for purposes of determining if any subsequent running of this same plasma process ABC in thissame chamber36 has proceeded in accordance with the spectral data from the plasma process ABC previously recorded in thenormal spectra subdirectory288.

Errors or aberrations which have been previously encountered while running a plasma process in theprocessing chamber36 are contained within a subdirectory of theplasma spectra directory284 of FIG. 9 which is entitled “Spectra of “Abnormal”Plasma Processes” and identified by reference numeral292 (hereafter “abnormal spectra subdirectory292”). Data relating to any of the plasma processes referenced above in relation to thenormal spectra subdirectory288 may also be stored in theabnormal spectra subdirectory292, and the above-noted organizational techniques may be utilized here as well. Entries to theabnormal spectra subdirectory292 are made when a given plasma process conducted in theprocessing chamber36 does not proceed in the desired or predetermined manner (e.g., when the process has not proceeded according to the relevant plasma process(es) of the normal spectra subdirectory288), and further when the cause or causes of the error or aberration has been identified to theplasma spectra directory284. This typically requires analysis of at least that portion of the spectral data from the time when the error or aberration in the subject plasma process first occurred, as well as possibly spectral data from the remainder of the subject plasma process. Errors or aberrations in a plasma process will typically be evident in the spectra of the plasma in theprocessing chamber36. Moreover, by obtaining data on the current plasma process in the Preferred Optical Bandwidth, at the Preferred Data Resolution, and at the Preferred Data Collection Time Resolution, the chances of obtaining optical emissions data which is indicative of the subject error or aberration is enhanced.

When spectral data on a current plasma process being conducted in theprocessing chamber36 deviates from the corresponding spectral data in thenormal spectra subdirectory288 in the determination of the currentplasma process module250, themodule250 may then compare this “deviating spectral data” on the current plasma process with spectral data in theabnormal spectra subdirectory292. Any number of actions may be initiated if the currentplasma process module250 identifies a “match” between the spectral data from the plasma process currently being conducted in theprocessing chamber36 and spectral data in theabnormal spectra subdirectory292. These actions may include issuing an appropriate alert(s) of the error condition, addressing one or more aspects of or relating to the control of thechamber36, or both as will be discussed in more detail below in relation to theprocess alert module428 of FIG.14.

Spectral data from a plasma process which is currently being conducted in thesubject processing chamber36, which does not “match” any plasma process stored within thenormal spectra subdirectory288, and which further does not “match” with corresponding spectral data in theabnormal spectra subdirectory292, is recorded in a subdirectory of theplasma spectra directory284 which is entitled “Spectra of “Unknown” Plasma Processes” and identified by reference numeral296 (hereafter “unknown spectra subdirectory296”). There are a number of circumstances when data from a plasma process which is being currently conducted in theprocessing chamber36 will be recorded in theunknown spectra subdirectory296. Any error or aberration which is “new” to the current plasma process module250 (i.e., spectral data, indicative of an error or aberration, which has not been previously recorded in the abnormal spectra subdirectory292) will result in relevant data from the current plasma process being recorded in theunknown spectra subdirectory296. Another circumstance where data may be entered in theunknown spectra subdirectory296 is when the current plasma process being conducted in theprocessing chamber36 is actually a new plasma process in relation to the currentplasma process module250. That is, the subject plasma process may very well be proceeding in accordance with the desired or predetermined manner, but data on this particular plasma process has not have been previously recorded in thenormal spectra subdirectory288 of FIG.9. As such, the spectral data of the current plasma process would not match any plasma process in thenormal spectra subdirectory288, and should not match any corresponding spectral data in theabnormal spectra subdirectory292. When an “unknown” condition is encountered during operations, an appropriate alert may be issued, control of the current process may be addressed, or both.

Spectral data recorded in theunknown spectra subdirectory296 from prior plasma processes will typically be analyzed by personnel at some point in time after the process has been terminated. If the spectral data from a plasma process recorded in theunknown spectra subdirectory296 is identified as being a new plasma process, and if a determination is made to use this spectral data as a standard for evaluating further runnings of this same plasma process on thissame processing chamber36, this spectral data may be transferred to thenormal spectra subdirectory292. Entries may also be made to theabnormal spectra subdirectory292 from theunknown spectra subdirectory296. Analysis of the spectral data from a particular plasma process which is recorded in theunknown spectra subdirectory296 may lead to the conclusion that the spectral data is associated with one or more particular errors/aberrations which is identifiable by its spectral data. The relevant spectral data from theunknown spectra subdirectory296 may then be transferred to theabnormal spectra subdirectory292.

Theplasma spectra directory284 of FIG. 9 also contains data which is indicative of when the endpoint has been reached of an entire plasma process or a discernible portion thereof such as a plasma step of a single or multi-step plasma recipe or other plasma process. “Endpoint” in the context of a plasma process or a discernible portion thereof is that time in the plasma process when the plasma within theprocessing chamber36 has achieved a certain predetermined result. Each plasma step in a plasma recipe typically has one or more characteristics in its corresponding spectra which will indicate that the desired predetermined result has been achieved, as typically does the end of a plasma clean which was initiated without first wet cleaning thechamber36, a plasma clean which was initiated after wet cleaning thechamber36, and a conditioning wafer operation. Spectral data of a plasma process conducted in achamber36 may be analyzed after the plasma process is terminated to identify one or more spectra (or portions thereof such as one or more individual wavelengths) which are indicative that endpoint of the subject plasma process or plasma process step has been reached. Spectral data which is indicative of endpoint from the various above-noted categories of plasma processes may be stored in a subdirectory of theplasma spectra directory284 of FIG. 9 which is entitled “Spectra of Endpoint Characteristic(s)” and identified by reference numeral316 (hereafter “endpoint subdirectory316”). The currentplasma process module250 may use the information contained in theendpoint subdirectory316 to issue an appropriate alert(s) of the identification of an endpoint condition, to address one or more aspects of or relating to the control of thechamber36, or both.

Multiple runnings of plasma processes within thechamber36 may “age” thechamber36 due to the nature of the plasma processes, and this aging may adversely affect the performance of thechamber36 in some manner. Indications that thechamber36 may be in need of some type of cleaning may be reflected by the spectra of the plasma in thechamber36. Spectral data may be included in theplasma spectra directory284 of FIG. 9 which corresponds with a condition where “cleaning” of the interior of theprocessing chamber36 would be desirable. “Dirty chamber condition” spectral data may be recorded in theabnormal spectra subdirectory292, in which case the “dirty chamber condition” would then be characterized as a known error or aberration consistent with the discussion above on theabnormal subdirectory292. Alternatively, a separate subdirectory may be employed as illustrated in FIG. 9 in the nature of a “Spectra of “Dirty Chamber Conditions” Subdirectory” and identified by reference numeral300 (hereafter “chamber condition subdirectory300”). The currentplasma process module250 may use this information on “dirty chamber conditions” to identify when aprocessing chamber36 is in condition for cleaning, and further such that appropriate actions may thereafter be undertaken.

A final subdirectory of theplasma spectra directory284 of FIG. 9 is a calibrationlight spectra subdirectory310 which does not contain spectra of plasma from thechamber36. Instead, one or more spectra of one or more calibration lights are stored in thesubdirectory310. Generally, a calibration light, whose spectra is in thecalibration light subdirectory310, is directed at thewindow38 of theprocessing chamber36. A comparision is made between the spectral pattern of the calibration light from thesubdirectory310 and the spectral pattern of that portion of the calibration light which is reflected by theinner surface40 of thewindow38 on theprocessing chamber36. The results of the comparison are used to determine the type and amount of calibration which should be implemented in relation to the operation of the currentplasma process module250 as will be discussed in more detail below in relation to thecalibration module562 and FIGS. 40-48.

The above-described structure of theplasma spectra directory284 and its various subdirectories presented in FIG. 9 is just that—a general structure which organizes categorically similar data for use by the currentplasma process module250 and its various sub-modules. The manner in which the data used by the currentplasma process module250 is actually stored is not particularly relevant for purposes of the present invention. However, it should be appreciated that data should be stored in a manner which allows the currentplasma process module250 to execute its monitoring/evaluation function in a timely fashion. Preferably, this is on at least a substantially “real-time” basis and sub-second acquisition, analysis, and control is available through themodule250. More specifically, acquisition of data, analysis of this same data, and initiation of a protocol(s) based upon this analysis may all be completed in less than a second through the currentplasma process module250.

Continuing to discuss the storage of spectral data used by the currentplasma process module250 in relation to the directory/subdirectory structure discussed above in relation to FIG. 9, one embodiment of how data may be stored in this directory/subdirectory structure is illustrated in FIG.10. Theplasma spectra directory284 may have the same subdirectories as presented in FIG. 9, although only thenormal spectra subdirectory288 andabnormal spectra subdirectory292 are illustrated for convenience and only data on one category of plasma process which may be stored in thenormal subdirectory288 is illustrated (plasma recipes). Review of thenormal spectra subdirectory288 of FIG. 10 indicates that spectra on multiple plasma recipes, which have been previously run on product in theprocessing chamber36 which is associated with theplasma spectra directory284, are each stored in their ownmain data entry350. This would also be the case with other categories of plasma processes stored in thenormal subdirectory288. That is, eachmain data entry350 is reserved for storing information which is used to evaluate plasma processes which are to be conducted in thissame processing chamber36.

Eachmain data entry350 for a given plasma process has a plurality of data segments354 associated therewith, and each of these data segments354 will include at least a spectra (e.g., FIG. 8) of the plasma in theprocessing chamber36 at a certain point in time and preferably within the Preferred Optical Bandwidth at the Preferred Data Resolution. The spectra associated with each data segment354 may be stored as a single spectra which covers the Preferred Optical Bandwidth, or may be stored as multiple spectra which collectively cover the Preferred Optical Bandwidth. Spectra for the data segments354 are taken periodically throughout the running of a plasma process within theprocessing chamber36 through thewindow38 on the chamber36 (e.g., by theplasma monitoring assembly174 of FIG. 6 or any of the embodiments illustrated in FIGS. 31 and 37 below) using the Preferred Data Collection Time Resolution. Although the entirety of the plasma process may be recorded in thenormal spectra subdirectory288 in this manner, sometimes the plasma is rather unstable when it first comes on in thechamber36. Therefore, it may be desirable to not retain optical emissions data in thenormal spectra subdirectory288 from this unstable time period.

Entries of plasma processes in thenormal spectra subdirectory288 may consist of a plurality of totally different types or species of plasma processes within a given category or genus as also illustrated in FIG.10. Plasma recipe A is stored undermain data entry350a,which is different from a plasma recipe B which is stored undermain data entry350b,which is different from a plasma recipe “X” which is stored undermain data entry350c.Multiple runnings of the same plasma recipe or process may also be recorded in thenormal spectra subdirectory288 as well if desired (not shown). For instance, spectral data from two separate runnings of plasma recipe A on the same type of product in the associatedprocessing chamber36 may actually be included in thenormal spectra subdirectory288. Evaluation of a current plasma recipe being run on product in thesubject processing chamber36 would then potentially involve the comparison of optical emissions data on the current process in relation to both of thesemain data entries350.

The optical emissions data within thenormal spectra subdirectory288 of FIG. 10 may be consolidated or condensed to eliminate the storage of redundant data, to increase the speed of the search of thenormal spectra subdirectory288 by the currentplasma process module250, or both. FIG. 11 illustrates one way in which this may be accomplished in the case of thenormal spectra subdirectory288afor one example where Plasma Recipes A-D are stored in thedirectory288a.The same principles would apply to any type of plasma process which is stored in thenormal spectra subdirectory288.

Plasma Recipe A undermain data entry358aand Plasma Recipe B undermain data entry358beach have the same spectra for purposes of the currentplasma process module250 from time t₁(the first time data is recorded in thesubdirectory288afor the subject plasma recipe) to time t_n(the “nth” time data is recorded in thesubdirectory288afor the subject plasma recipe). Instead of storing the spectra over this time range twice in thenormal spectra subdirectory288a(once undermain data entry358afor Plasma Recipe A and once undermain data entry358bfor Plasma Recipe B), the multiple spectra from this time range are stored only once in common data segments362a-c.More than 3 common data segments362 may obviously be used. Nonetheless, common data segments362a-care thereby associated with both Plasma Recipe A ofmain data entry358aand Plasma Recipe B ofmain data entry358b.At time t_n+1(i.e., the first time data is recorded in thenormal spectra directory288aafter the time t_n) and until the end of the plasma recipe in the example presented by FIG. 11, however, the spectra of Plasma Recipe A undermain data entry358aand the spectra Plasma Recipe B undermain data entry358bdiffer for purposes of the currentplasma process module250. As such, Plasma Recipe A undermain data entry358aand Plasma Recipe B undermain data entry358beach then include their own individual data segments366a-cand366d-f,respectively, over the time period from t_n+1to t_n+x(the “xth” time data is recorded in thenormal spectra subdirectory288aafter the time t_n). Each of the Plasma Recipes A and B end at the same time for purposes of the example of FIG. 11 (i.e., at time t_n+x), although in a commercial setting this may not be the case.

The normal spectra subdirectory288bof FIG. 11 also hasmain data entries358cand358dfor Plasma Recipes C and D, respectively. The same data storage concept used in relation to Plasma Recipes A and B is likewise employed for Plasma Recipes C and D in thenormal spectra subdirectory288a.The spectra of Plasma Recipe C under datamain entry358cand Plasma Recipe D undermain data entry358dare the same for purposes of the currentplasma process module250 from time t₁to time t_n. Instead of storing the spectra over this time range twice in the normal spectra subdirectory288b(once undermain data entry358cfor Plasma Recipe C and once undermain data entry358dfor Plasma Recipe B), the multiple spectra from this time range are stored only once in common data segments362d-hin thenormal spectra subdirectory288a.Common data segments362d-hare thereby associated with both Plasma Recipe C ofmain data entry358cand Plasma Recipe D ofmain data entry358d.At time t_nand until the time t_n+xof the plasma recipe in the example presented by FIG. 11, however, the spectra of Plasma Recipe C undermain data entry358cand the spectra of Plasma Recipe D undermain data entry358ddiffer for purposes of the currentplasma process module250. As such, Plasma Recipe C undermain data entry358chas its own individual data segments366g-kover the time period from t_n+1to t_n+x, while Plasma Recipe D undermain data entry358din turn has its own individual data segments366l-pover this same time period. At time t_n+xand until the end of the plasma recipe in the example presented by FIG. 11 at t_n+y, however, the spectra of Plasma Recipe C undermain data entry358cand the spectra of Plasma Recipe D undermain data entry358dagain are the same for purposes of the currentplasma process module250. As such, Plasma Recipe C undermain data entry358cand Plasma Recipe D undermain data entry358dhave common data segments362i-zover the time period from t_n+xto t_n+y. Each of the Plasma Recipes C and D end at the same time for purposes of the example of FIG. 11, although in a commercial setting this again may not be the case.

Each data segment354 of each plasma process stored under amain data entry350 in thenormal spectra subdirectory288 of FIG. 10 may contain a multiplicity of data types relevant to the monitoring of the current plasma process with the currentplasma process module250. A representative example is presented in FIG. 12A where these various data types of data are presented in data fields322 which are associated with each data segment354. Spectral patterns of the plasma in theprocessing chamber36 is a significant data type for comparing the current plasma process with theplasma spectra directory284, and these spectra are stored in aspectra field322din thenormal spectra subdirectory288 of FIG.12A. Each data segment354 in thenormal spectra subdirectory288 also includes atime field322awhere the time associated with the spectra in thespectra field322dis recorded (e.g., the time into the plasma process when the spectra is taken). Data in thetime field322amay be used in various ways by the currentplasma process module250 as will be discussed in more detail below.

Further information is at least associated with eachmain data entry350 in thenormal spectra subdirectory288. “Associated” in this context means that this information may be provided once for eachmain data entry350 for a given plasma process or for only a few times which is less than the total number of data segments354 under a particularmain data entry350, but it also encompasses a situation where this information is actually provided for each data segment354 of the subjectmain data entry350 which is not as desirable because of redundancies. Fields for these types of information include: a plasma process “genus”field322h(e.g., to identify whether themain data entry350 is a plasma recipe, plasma clean, or conditioning wafer operation); awafer identifier field322b(e.g., for information which corresponds with an identifier, such as a number or code, which appears on thewafer18 which is to have a plasma recipe run thereon and which is used for tracking purposes); a plasma process “species”field322c(a subset of a plasma process “genus”, such as different types of plasma recipes (e.g., plasma recipe A and plasma recipe B)); a plasmaprocess step field322e(e.g., to identify the step of a particular plasma recipe or any other plasma process which provides a different function or achieves a different result than other portions of the process); a maximum total plasma processstep time field322f(e.g., the maximum amount of time allowed to complete a given plasma step of a multi-step plasma recipe or other process); and a maximum total plasmaprocess time field322g(e.g., the maximum amount of time allowed to complete an entire plasma process (each of its steps)). Certain of this information will be inapplicable to certain genus of plasma processes and/or certain species of plasma processes within a certain plasma process genus. Information provided to the above-noted fields may be input with thedata entry device132 of FIG. 6, although the information for thewafer identifier field322bmay be automatically read from thewafer18 and input to the subject main data entry354 (e.g., scanner).

Theplasma spectra directory284 of FIG. 10 also illustrates one way of storing data for entries in theabnormal spectra subdirectory292. A review of theabnormal spectra subdirectory292 of FIG. 10 indicates that spectral data on multiple known errors or aberrations are each stored in a main data entry346 for evaluating future plasma processes which are conducted in thesame processing chamber36 where these errors or aberrations occurred. As noted above, these errors in the main data entries346 have preferably been identified (e.g., the cause(s) of the error has been determined), and are thereby a “known” condition which may be encountered when running a plasma process in theprocessing chamber36. Each main data entry346 having one specific error therein is illustrated as having a plurality of data segments354 associated therewith, and each of these data segments354 will include at least a spectra (e.g., FIG. 8) of the plasma from theprocessing chamber36 which is a deviation from the relevant spectra of the corresponding plasma process in thenormal spectra subdirectory288 and which is indicative of the subject error or aberration. The spectra associated with each data segment354 under theabnormal spectra subdirectory292 may be of the Preferred Optical Bandwidth using the Preferred Data Resolution. Alternatively, the spectra associated with the data segments354 under theabnormal spectra subdirectory292 may only include that portion of the spectrum which contains the characteristic(s) which are indicative of the error in question (i.e., an optical emissions segment which is contained within, but smaller than, the Preferred Optical Bandwidth).

Further information relevant to theabnormal spectra subdirectory292 is illustrated in FIG.10. Spectral patterns for the data segments354 under each main data entry346 of theabnormal spectra subdirectory292 are illustrated as having been recorded periodically during the running of the entire plasma process within the processing chamber36 (e.g., using the Preferred Data Collection Time Resolution). Data on the entire plasma process could be retained in theabnormal subdirectory292. In this case the time t₁referenced in thedata segments354m,354q,and354uwould be the first spectra obtained of the subject plasma process, while the time t_nwould be the last spectra obtained of the subject plasma process at the termination thereof (e.g., endpoint of the last step in the process, or the endpoint of the plasma process if there is not more than one step). This situation may result in the storage of unnecessary data in theabnormal spectra subdirectory292. Consider a situation where a plasma recipe is being run on product within theprocessing chamber36 and has proceeded according to thenormal spectra subdirectory288 for the first 90 seconds of the plasma recipe. Assume that only about 10 additional seconds are required to complete the plasma recipe (i.e., to achieve the result associated with the last step of the recipe). Also assume that an error occurs at the 91 second mark in the current plasma recipe. Unless the data on the first 90 seconds of the plasma recipe provides some type of indication as to the upcoming error which again occurred at the 91 second mark in the current plasma recipe of this example, this data will not be useful for retention in theabnormal spectra subdirectory292. In this case the spectra at the time t₉₁could be the first spectra which did not “match” with any of the plasma recipes in thenormal spectra subdirectory288. No other spectra need be included under the main entry346 if this one spectra sufficiently identifies the error (not shown). However, it may be desirable to record in theabnormal spectra subdirectory292 all spectra which are obtained after the error has been identified and up until the plasma process is terminated at the time t_n, or at least a few spectra at various time intervals. Flexibility is provided by theabnormal subdirectory292 in that the amount of data included under each of the main data entries346 in which the errors are contained may be independently selected. Therefore, the data segments354 under each main data entry346 of theabnormal spectra subdirectory292 may not necessarily provide a complete history of a plasma process in which an error occurred.

Multiple data types may be included in each data segment354 associated with each error in each main data entry346 of theabnormal spectra subdirectory292 of FIG. 10. A representative example is presented in FIG. 12B where these various data types are contained within data fields338 which are associated with each data segment354 under each main data entry346 of thesubdirectory292. Spectral patterns of the plasma in theprocessing chamber36 are a significant data type for comparing the current plasma process with theplasma spectra directory284, and these spectra are stored in aspectra field338bof theabnormal spectra subdirectory292 of FIG. 12B of each data segment354. Moreover, the category or genus of the subject plasma process may be identified in a plasmaprocess genus field338e(e.g., plasma recipe, plasma clean, conditioning wafer operation), a particular type or species of a given category or genus of plasma process may be identified in a plasma process species field338f(e.g., a particular type of plasma recipe), and the type of plasma step may be identified in a plasmaprocess step field338g.

Data in addition to the above-described spectra may be associated with each of the data segments354 under each main data entry346 for known error/aberrations within theabnormal spectra subdirectory292. Each data segment354 of each main data entry346 may also include atime field338afor containing information on the time which is associated with the spectra in the subject data segment354. Other information associated with the data segments354 under each main data entry346 for known error/aberrations within theabnormal spectra subdirectory292 includes information which identifies the error. Any textual identification or description of the error stored under a main data entry346 may be included in anerror field338cof theabnormal spectra subdirectory292. This information would typically be manually entered by personnel using the data entry device132 (e.g., FIG. 6) after the spectra was analyzed and the cause(s) of the error(s)/aberration(s) was identified.

The currentplasma process module250 includes error identification capabilities as will be discussed in more detail below. Once the currentplasma process module250 identifies a match between the current optical emissions data and a relevant spectra or portion thereof in theabnormal spectra subdirectory292, information on the corresponding error/aberration may be issued based upon information in theerror field338c.Moreover, corrective actions may be undertaken based upon the contents of thissame error field338c.In this regard, each data segment354 under each main data entry346 for known error/aberrations within theabnormal spectra subdirectory292 may also include aprotocol field338d.Information contained within the protocol field33dwill somehow relate to how the subject error or aberration may or should be addressed, or more specifically what action or actions may or should be undertaken to address the error. Single or multiple protocols may be stored in any oneprotocol field338d(e.g., more than one protocol may be appropriate to address a certain condition). Once the currentplasma process module250 identifies a match between the current spectra and a relevant spectra in theabnormal spectra subdirectory292, how the corresponding error/aberration is addressed may be based upon information contained in thesubject protocol field338d.

Data regarding process control parameters or conditions associated with theprocessing chamber36 may also be included in each of the above-described subdirectories of theplasma spectra directory284 for a specific data entry, particularly in the case of theabnormal spectra subdirectory292 and the unknown conditions subdirectory296. These would include conditions which are typically monitored in a plasma process, such as the types of feed gases being used or the composition of the plasma, temperatures within one or more regions of thechamber36, the pressure within theprocessing chamber36, power settings, and gas flow rates.

Pattern Recognition Module370—FIG.13

Certain key principles of the present invention are simply based on whether one spectral pattern matches another spectral pattern (e.g., does the spectral pattern of the plasma in thechamber36 “match” the pattern of the relevant spectra in the relevant subdirectory of the plasma spectra directory288). In many cases this determination may be made through thepattern recognition module370 which is presented in FIG.13. Various “pattern recognition techniques” may be employed by thepattern recognition module370 to provide the above-noted function. One such pattern recognition technique is embodied by the flowchart depicted in FIG.13 and may be generally characterized as a point-by-point pattern recognition technique. The point-by-point pattern recognition technique which is embodied by thepattern recognition subroutine374 of FIG. 13 is contained withinstep378. The intensity at a first wavelength in the current spectra at the current time t_c(a fixed point in time) is compared with the intensity at this same first wavelength of the relevant spectra from a Target Directory. Whichever sub-module of the currentplasma process module250 calls thepattern recognition module370 will designate which particular subdirectory of theplasma spectra directory284 is to be searched by themodule370 for “matching” patterns (and thereby defines the Target Directory). The sub-module which calls thepattern recognition module370 may also establish what will constitute a “matching” pattern. That is, what may be a “matching spectra” in relation to one sub-module of the currentplasma process module250 may not be a “matching spectra” in relation to another of its sub-modules.

If the intensities of the two subject spectra are within a “match limit” of each other at this first wavelength in the subject optical emissions, the patterns of the two subject spectras are initially considered a “match” and the analysis is repeated at a second wavelength which is displaced and different from the first wavelength and which defines a second “point” on which the above-noted point-by-point analysis is repeated. The particular “match limit” which is used by thepattern recognition subroutine374 may be specific to which sub-module of the currentplasma process module250 calls thepattern recognition subroutine374 as noted. The above-described point-by-point analysis is repeated by “advancing along” the entirety of the current spectra at typically pre-selected wavelength increments (e.g., every nanometer). Typically, a fixed wavelength increment will be utilized bystep378, such that the comparison between the two subject spectra will be made at every “x” nanometers throughout the entirety of the entire “bandwidth” of the spectra. However, there need not be equal spacings between each of the “points” examined by thepattern recognition subroutine374.

Another match criterion which may be used in combination with the above-noted point-by-point comparison relates to how many of the examined points must be within the “match limit” in order for the two subject spectra to be considered a match. The sub-module which calls thepattern recognition subroutine374 may require each “point” examined in the subject point-by-point analysis to be within the selected match limit in order for the two spectra to be considered a match. Alternatively, something less than 100% may also be utilized. For instance, having at least 95% of the points being within the match limit may equate with the two subject spectras being considered a match. Furthermore, the average variation of the multiplicity of points examined by thepattern recognition subroutine374 may be calculated and compared with an average associated with the relevant spectra from the Target Directory to determine if it is within a predefined tolerance. Any combination of the foregoing may be implemented for determining what is a “match.”

A number of factors will affect the accuracy which may be attributed to the results achieved by the execution ofstep378 in thepattern recognition subroutine374 of FIG.13. One such factor is the “match limit” which will be discussed in relation to the point-by-point comparison technique, but which will be equally applicable to the averaging discussed above. In this case, the “match limit” is the amount that the intensity associated with a point in the current spectra at a certain point in time t_cwill be allowed to deviate from the intensity of the relevant spectra from the Target Directory, and yet still be considered a “match” for purposes of thepattern recognition subroutine374. Two types of “match limits” which may be utilized include a raw difference basis and a percentage difference basis. In the case of a “raw difference basis”, a fixed number of intensity units is established and input to thepattern recognition subroutine374 and which thereby defines the boundary of a “match” (e.g., the “raw difference basis” limit may be established at ±“x” intensity units where “x” is a value which is input to thepattern recognition subroutine374, such that the intensity at each wavelength of the current spectra checked by thepattern recognition subroutine374 must be within ±“x” intensity units of the intensity at each of the corresponding wavelengths of the relevant spectra from the Target Directory in order to be considered a “match”). In the case of having a “match limit” based upon a “percentage difference basis,” a fixed percentage is established and input to thepattern recognition subroutine374 and which thereby defines the boundary of a “match” between corresponding intensities of the current spectra and the relevant spectra from the Target Directory (e.g., the “raw difference basis” limit may be established at ±“x” percent, such that the intensity at each wavelength of the current spectra checked by thepattern recognition subroutine374 must be within ±“x” percent of the intensity at each of the corresponding wavelengths of the relevant spectra from the Target Directory in order to be considered a “match”). Both the “raw difference” and “percentage difference” may be used simultaneously as a “match limit” by thepattern recognition subroutine374 as well (i.e., both criterion must be satisfied in order for the spectras to be considered a “match”). The “match limits” are equally applicable to when the above-noted averaging technique is used as well. Regardless of which type of “match limit” is employed, it is entered for use by thepattern recognition subroutine374.

Another factor which affects the accuracy which may be attributable to the results of thepattern recognition subroutine374 of FIG. 13 is the analytical wavelength resolution used in the above-described point-by-point analysis. “Analytical Wavelength Resolution” in this context means the wavelength increments at which the above-described point-by-point analysis will be performed throughout the subject spectra. Although the plurality of wavelengths for the above-noted point-by-point analysis could be random throughout the bandwidth of the subject spectra, preferably a pattern is used such as a fixed wavelength increment. For instance, if the analytical wavelength resolution is established at 1 nanometer and input to thepattern recognition subroutine374, the above-noted point-by-point analysis ofstep378 would be performed at each 1 nanometer increment across the entirety of the Preferred Optical Bandwidth. Preferably the Analytical Wavelength Resolution used by thepattern recognition subroutine374 is no more than about 2 nanometers, and even more preferably no more than about 0.5 nanometers. Hereafter, this will be referred to as the “Preferred Analytical Wavelength Resolution.”

Summarizingstep378 of thepattern recognition subroutine374 of FIG. 13 in the case where the current spectra extends from 200 nanometers to 900 nanometers and where the Analytical Wavelength Resolution is 1 nanometer, a comparison is made of the intensities at the 200 nanometer wavelength from the current spectra at the current time t_c(from the current plasma process being run within the processing chamber36) and from the relevant spectra from the Target Directory. If the difference between these two intensities is within the “match limit” input to thepattern recognition subroutine374, whether using a raw difference theory, a percentage difference theory, or a combination of a raw difference theory and a percentage difference theory (e.g., the difference in intensity must be no more than a certain, input number of intensity units and must also be within a certain, input number of percentage points of each other in the case of a combination), the 200 nanometer wavelength “point” of the current spectra and the relevant spectra from the Target Directory will be characterized as a “match” at the current time t_c. The point-by-point analysis will then continue at the 201 nanometer wavelength in the above-described manner, and will be repeated at each 1 nanometer increment until reaching the last 900 nanometer wavelength. Results of the point-by-point comparison ofstep378 for the current spectra at the current time t_cwill then be provided to step380 of thepattern recognition subroutine374 of FIG. 13 for use by the sub-module of the currentplasma process module250 which called thepattern recognition module370. Control of the plasma monitoring operations is then returned bystep382 of thepattern recognition subroutine374 to the sub-module of the currentplasma process module250 which called thepattern recognition module370.

Some plasma processes change very rapidly and some plasma processes are of relatively short duration (e.g., some plasma steps of a plasma recipe are no more than about 5 seconds). Therefore, spectral data should be taken at least every 1 second and the analysis of this data should be completed by thepattern recognition subroutine374 as fast as possible. In the case where the plasma process is a plasma recipe which will be discussed in more detail below in relation to theplasma health subroutine253 of FIG. 21, the identification of the current plasma recipe and the analysis of the performance of the processing chamber36 (e.g., plasma health) should be completed at least before thenext wafer18 is loaded into thechamber36 for another running of a plasma recipe on this product within thechamber36. Thepattern recognition subroutine374 of FIG. 13 is able to meet the demands through simplifying the analysis of the spectra of the plasma in theprocessing chamber36. The sum total of the analysis provided by thepattern recognition subroutine374 is simply whether the pattern of the current spectra “matches” the pattern of the relevant spectra from the Target Directory. There is no need to locate or define peaks in the spectra of the plasma from the process currently being conducted in theprocessing chamber36 in the analysis used by thepattern recognition subroutine374 at each of the limits. Nor is any attempt made by thepattern recognition subroutine374 to identify the various chemical species currently in the plasma in theprocessing chamber36 through spectral analysis. Again, the only determination made by thesubroutine374 is whether the current spectral pattern “matches” the relevant spectral pattern in the Target Directory. In one embodiment thepattern recognition subroutine374 is able to executestep378 for a spectra defined by wavelengths at least within the range of about 150 nanometers to about 1,200 nanometers with an analytical wavelength resolution of no more than about 1 nanometer (i.e., execute the point-by-point analysis at least at every 1 nanometer increment) in no more than about 1 second, and preferably in no more than about 0.5 seconds.

Process Alert Module428—FIG.14

Various conditions which may be encountered by the currentplasma process module250 may result in the transfer of control to or the sharing of control with theprocess alert module428 of FIG.14. One or more subroutines may be included under theprocess alert module428. Each of these subroutines may present various options in relation to how the relevant condition or situation is addressed which resulted in the activation of theprocess alert module428. In the case of theprocess alert subroutine432 of FIG. 14, two categories of “actions” are made available—the issuing of one or more alerts and addressing the control of the subject plasma process in some manner.

One or more alarms, alerts, or the like may be activated if the alarm alert function of theprocess alert subroutine432 of FIG. 14 is enabled at itsstep454 in relation to the subject condition or situation. At least one visual alarm may be activated instep458 of theprocess alert subroutine432. Exemplary visual alarms include displaying a general indication of the existence of the relevant condition (e.g., a flashing light), a more specific indication of the subject condition (e.g., providing a textual description of the identified condition or situation), or both. Appropriate locations where information relating to the subject condition may be presented include thedisplay130 associated with theparticular processing chamber36 where the subject condition was encountered, thedisplay59 associated with thewafer production system2 which may be characterized as a master control panel of sorts for thewafer production system2, any master control panel for the entire fabrication facility incorporating thewafer production system2, any computer network on which thewafer production system2 is included, or on any combination of the foregoing. Other visual indications may be employed alone or in combination with any of the foregoing. Audio and any other types of alarms may also be employed.

Another option available under theprocess alert subroutine432 of FIG. 14 relates at least in some manner to the control of the plasma process and which may be accessed throughstep436 of theprocess alert subroutine432. Any spectra which is associated with a condition, or the identification of the condition itself, which will trigger the activation of theprocess alert subroutine432 may be included in or associated with a step448 of thesubroutine432. One or more protocols which have been established for when this spectra or condition is encountered in thechamber36 may be included in an associatedstep450. Multiple spectra or conditions may be included in any one step448. The commonality between these spectra/conditions included in or associated with a step448 is the categorically similar protocol included in its associatedstep450.

Five categorical protocols are presented in FIG.14. Step450apresents a protocol category which is the modification of one or more process control parameters. One or more spectra from theabnormal spectra subdirectory292, one or more conditions, or both, may be included instep448aof theprocess alert subroutine432 which will access step450a.Step450ais directed toward a protocol category which is to attempt to “address” (e.g., correct/remedy) the subject condition in the current plasma process being run in thechamber36. The protocol associated withstep450aof theprocess alert subroutine432 more specifically provides for the modification or adjustment of one or more process control parameters in a manner which has been previously determined to be appropriate to address the subject condition. Adjustment of the process control parameters associated with the current plasma process may be automatically undertaken if desired by the facility incorporating thewafer production system2 through operatively interfacing theprocess alert subroutine432 with the appropriate process controller(s) (e.g., by sending an appropriate signal from thePMCU128 to theMCU58 of FIG. 1 which controls the wafer production system2). Manual adjustment of one or more process control parameters is also contemplated bystep450aof theprocess alert subroutine432. Execution ofstep450ain this case would entail informing the appropriate personnel of the protocol(s) associated with the subject condition such that these personnel could manually initiate the appropriate action if desired.

Notwithstanding the presentation of a single “modify process control parameters” protocol in FIG. 14, it should be appreciated that different process control modifications may be initiated for different spectra/conditions associated withstep448a.One or more spectra may be associated with a condition which requires modification of the process control parameters in one way, while one or more spectra associated with another condition may require modification of the process control parameters in another way. Moreover, any one or more spectra or one or more conditions associated withstep448amay have one or more process control protocols associated therewith. For instance, in the case wherestep450ais not directly integrated with the relevant process controller(s), a listing of possible corrective actions which may be undertaken to address the associated condition(s) may be provided for consideration by the appropriate personnel. Integratingstep450awith one or more controllers associated with thewafer production system2 may still utilize multiple process control protocols for any one or more spectra/conditions. Attempts to address the subject condition associated withstep448amay first be pursued through a first protocol associated with this condition and step450a.If this is unsuccessful in addressing the subject condition, a second protocol associated with the subject condition and step450amay then be undertaken and so forth.

Another categorical protocol which may be included in theprocess alert subroutine432 of FIG. 14 relates to terminating the current plasma process. One or more spectra which are representative of one or more conditions, or the identification of the condition(s) itself, may be included in or associated withstep448bof thesubroutine432 which accessesstep450b.Step450bcontains one or more protocols which are directed toward terminating the subject plasma process, although typically termination of the current plasma process will simply entail terminating the gas flow to thechamber36 and the electrical componentry which are responsible for creating the plasma. Termination of the current plasma process may be automatically undertaken if desired by the facility incorporating thewafer production system2 through operatively interfacing theprocess alert subroutine432 with the appropriate process controller(s) (e.g., by thePMCU128 sending an appropriate signal to the MCU58). Manual termination of the current plasma process is also contemplated bystep450b.Execution ofstep450bin this case may then simply entail apprising the appropriate personnel that a condition has been identified where termination of the current plasma process being run in theprocessing chamber36 is recommended such that the appropriate action may be manually undertaken if desired.

Cleaning operations may also be initiated through theprocess alert subroutine432 of FIG.14. One or more spectra from theabnormal spectra subdirectory292 or thechamber condition subdirectory300, or again simply the identification of the relevant condition, may be included in/associated withstep448eof thesubroutine432 which in turn accesses step450e.Step450econtains a protocol which is directed toward initiating some type of a cleaning of the interior of theprocessing chamber36. Cleaning operations may be automatically undertaken if desired by the facility incorporating thewafer production system2 through operatively interfacing theprocess alert subroutine432 with the appropriate process controller(s) (e.g., by thePMCU128 sending an appropriate signal to the MCU58). Manual implementation of these actions is also contemplated bystep450e.In this regard, the execution ofstep450ein theprocess alert subroutine432 of FIG. 14 may simply entail apprising personnel that it is recommended that the current plasma process being run in theprocessing chamber36 be terminated due to the detected dirty chamber condition, and that a cleaning operation be subsequently manually initiated.

One or more spectra/conditions associated withstep448emay be associated with different protocols instep450e.For instance, one protocol ofstep450ecorresponding with one or more spectra/conditions associated withstep448emay relate to a plasma cleaning operation which may be initiated in accordance with the foregoing. Other spectra or conditions associated withstep448emay access a protocol ofstep450ewhich relates to a wet clean which may be initiated in accordance with the foregoing.

Spectra or conditions within thechamber36 which are of a nature such that the wafer distribution sequence should be affected in some manner by their existence may be included in or associated withstep448c.The protocol set forth instep450cthereby addresses the manner in whichwafers18 are distributed to thevarious processing chambers36 of thewafer production system2 through thewafer distribution module1384 which will be discussed in more detail below in relation to FIGS. 59-60. Addressing the sequence of distribution ofwafers18 to theprocessing chambers36 of thewafer production system2 may be automatically undertaken if desired by the facility incorporating thewafer production system2 through operatively interfacing theprocess alert subroutine432 with the appropriate process controller(s) (e.g.,wafer distribution module1384, MCU58). Manual techniques are also contemplated bystep450cin that the execution ofstep450cof theprocess alert subroutine432 may simply entail apprising personnel that it is recommended that the distribution sequence to thechambers36 of thesystem2 be manually addressed because of the existence of the subject condition.

Finally, plasma process/plasma process step endpoint may be addressed through theprocess alert subroutine432 of FIG.14. In this regard, one or more spectra indicative of the endpoint of the subject plasma process or discrete portion thereof (e.g., plasma process step), or the identification of the plasma process/process step itself, may be included in or associated withstep448d.The protocol set forth instep450daddresses how the identification of the occurrence of the particular endpoint should be addressed. This may include terminating the subject plasma process/process step, initiating the next plasma process/step (e.g., if the subject plasma step is not the last step of a given plasma recipe or other process), or both depending upon the nature of the plasma process. Automation and manual techniques are contemplated bystep450das in the above-noted cases.

The control of the plasma monitoring operations is relinquished by theprocess alert subroutine432 of FIG.14 through steps440 (if only the alarm alert function is enabled at step454) or462 (if the process control feature is enabled at step436). Depending upon the circumstances, control may be returned to the particular sub-module of the currentplasma process module250 which called theprocess alert module370. Another option which may be employed is to return control of the plasma monitoring operations in a particular case or in all cases to thestartup module202 of FIG.7 through execution ofsteps440 or462.

Startup Module202—FIGS.15-16

Information on what is happening in the processing chamber36 (e.g., spectra of the plasma in the chamber36) is made available to the currentplasma process module250 for evaluation of the current plasma process operation through its various “sub-modules” as generally discussed above and as will be addressed in more detail below in relation to the following relevant figures. Access to the various “sub-modules” of the currentplasma process module250 may be controlled through thestartup module202 of FIG.15. As such, thestartup module202 may be viewed as a main menu of sorts for the various options that are available through the currentplasma process module250.

One embodiment of a startup routine which may be used by thestartup module202 is illustrated in FIG.15 and provides the above-noted “main menu-like” function. Thestartup routine203 basically allows personnel to “enter” in some manner the type of action to be undertaken such that control of the plasma monitoring operations may be transferred to the appropriate sub-module of the currentplasma process module250. “Entry” may be accomplished by providing a listing on thedisplay130 associated with the PMCU128 (e.g., FIG. 6) of all of the actions which may be undertaken and allowing personnel to select which option should be pursued with thedata entry device132. Another option would be to allow personnel to input the action to be initiated using thedata entry device132. Still another option would be for thestartup routine203 to sequentially scroll down through a listing of the various options. Finally, no input need be provided as the currentplasma process module250 may just immediately begin comparing the current plasma process to the plasma spectra directory284 (e.g., using an appropriate order for searching the various subdirectories).

Three “categorical” actions may be initiated through thestartup routine203. First, certain calibration operations may be undertaken throughstep136 of thestartup routine203 which accesses acalibration module562 through execution ofstep140. Thecalibration module562 will be discussed in more detail below in relation to FIGS. 40-48. Research in relation to the current plasma process to be run in thechamber36 may be initiated throughstep144. For instance, research may be undertaken to identify one or more characteristics which are indicative of the endpoint of a particular plasma process or plasma process step. This is accomplished throughexecution step148 of thestartup routine203 which calls aresearch module1300 which will be discussed in more detail below in relation to FIGS. 49-51C.

A final option available through thestartup routine203 of FIG. 15 relates to current plasma processes (i.e., any plasma process run in thechamber36 which is not recorded in the plasma spectra directory284). Plasma processes such as plasma processing qualification/production wafers (step230), plasma cleaning operations without first doing a wet clean of the chamber36 (step234), plasma cleaning operations conducted after thechamber36 has been wet cleaned (step238), and conditioning wafer operations (step242) each may be accessed through thestartup routine203. The endpoint of these types of plasma processes, a specific portion thereof, or both may be determined through anendpoint detection module1200 which will be discussed in more detail below in relation to FIGS. 52-58 and which is called bystep240 of thestartup routine203. The “health” of these types of plasma processes may also be evaluated through theplasma health module252 which will be discussed in more detail below in relation to FIGS. 21-25 and which is called through execution ofstep236 of thestartup routine203.

Step236 of thestartup routine203 of FIG. 15 relates to a plasma health evaluation and calls thestartup subroutine204 of FIG.16. Two main options may be pursued in relation to “plasma health” through thestartup subroutine204 of FIG.16. Either a current plasma process may be recorded in thenormal spectra subdirectory288 to be used as a standard for evaluating plasma processes subsequently run in thechamber36, or the current plasma process may be evaluated against thenormal spectra subdirectory288. In this regard, step208 of thestartup subroutine204 of FIG. 16 inquires as to whether the plasma process to be conducted in thesubject processing chamber36 should be recorded in thenormal spectra subdirectory288 associated with thischamber36. If the “response” to the inquiry ofstep208 is a “yes”, thestartup subroutine204 proceeds to step224 where a determination is made as to the status of the plasma in theprocessing chamber36—specifically whether the plasma is “on” through optical analysis by the currentplasma process module250. One way to determine if the plasma is “on” in thechamber36 is to determine when the spectra obtained from theprocessing chamber36 “matches” any spectra stored in theplasma spectra directory284 or any of its subdirectories, such as through thepattern recognition module370 of FIG.13. Another way in which this may be done is to determine when any of the spectra from the interior of theprocessing chamber36 have at least a certain number of discrete peaks of a least a certain intensity. Using the same principles discussed above in relation to thepattern recognition module370 of FIG. 15 may also identify this type of spectra through the currentplasma process module250. Determining when there is at least a certain change in the optical emissions from within thechamber36 may also be indicative that the plasma is “on” (e.g., going from a “dark” condition to a “light” condition). Regardless of how the determination is made as to whether plasma exists in theprocessing chamber36, the “plasma on” indication may be appropriately conveyed to operations personnel or others in one or more of the above-noted manners.

Once plasma is present in theprocessing chamber36, thestartup subroutine204 proceeds to step228 where at least spectral data of the current plasma process is recorded in thenormal spectra subdirectory288. Preferably this encompasses the Preferred Optical Bandwidth at the Preferred Data Resolution and using the Preferred Data Collection Time Resolution. After the plasma process is terminated, thesubroutine204 returns to the “main menu-like”startup routine203 of FIG. 15 viastep226.

The other alternative available through thestartup subroutine204 of FIG. 16 is to evaluate the current plasma process to be run in thesubject processing chamber36 against spectral data already recorded in thenormal spectra subdirectory288. In the example presented in FIG. 16, this is accomplished by exitingstep208 of thestartup subroutine204 under a “no” logic condition, which directs thestartup subroutine204 to proceed to step212. Step212 inquires as to whether the plasma process to be run in theprocessing chamber36 should be evaluated against spectral data in thenormal spectra subdirectory288. Since the two main options available under thestartup subroutine204 are to either record data or evaluate spectral data of the current plasma process against spectral data already recorded in thenormal spectra subdirectory288, and further since thesubroutine204 reachedstep212 because a “decision” was made atstep208 of thesubroutine204 to not record data in thenormal spectra subdirectory288, a response of “no” atstep212 merely redirects thestartup subroutine204 to start over. Responding “yes” atstep212, however, directs thestartup subroutine204 to proceed fromstep212 to step216. Step216 inquires as to whether the plasma is on in theprocessing chamber36, and therefore may be identical to step224 discussed above. Once plasma exists within theprocessing chamber36, thestartup subroutine204 proceeds to220 where control of plasma monitoring operations is transferred to theplasma health module252 such that the health of the plasma process may be assessed. The “record” or “compare” options may be presented in other ways than as set forth in the start-upsubroutine203.

Plasma Health Evaluations

Theplasma health module252 of FIG. 7 is also included in the embodiment of FIG.32 and evaluates the overall health of the plasma in thesubject chamber36 or the “plasma health.”“Plasma health” as used herein means the state or the condition of the plasma process as it relates to plasma performance when compared to typical “normal” plasma behavior resulting in usable product. The “condition” of the plasma in turn may be characterized as the cumulative result of all parameters having an effect on the plasma in the processing chamber. Stated another way, “plasma health” may be equated with a condition where a current plasma process is proceeding in accordance with one or more plasma processes stored in thenormal spectra subdirectory288. In this regard, theplasma health module252 is able to determine if the current plasma process being conducted within theprocessing chamber36 is progressing “normally” through a comparison of at least a portion of the optical emissions from theprocessing chamber36 during the plasma process with the relevant spectra or portion thereof in thenormal spectra subdirectory288. Spectral patterns of the plasma in thechamber36 will change as the plasma process progresses. Moreover, the spectral patterns of the plasma in thechamber36 differ in relation to the category of plasma process being run. This is evidenced by a review of exemplary spectra from a plasma recipe, a plasma clean conducted without first wet cleaning thechamber36, a plasma clean executed after a wet clean of thechamber36, and a conditioning wafer operation presented below. In each case, “intensity” is plotted along the “y” axis and expressed in “counts” which is reflective of the intensity level, while “wavelength” is plotted along the “x” axis in nanometers.

Exemplary Plasma Recipe Spectra—FIGS.17A-C

An example of a multiple step plasma recipe run on awafer18 in thechamber36 is illustrated in FIGS. 17A-C in which the spectra of the plasma in theprocessing chamber36 varies with a change in the current plasma step. FIGS. 17A-C present aspectra744 of an exemplary first plasma step of an exemplary plasma recipe A, aspectra752 of an exemplary second plasma step of this same plasma recipe A, and aspectra760 of an exemplary third plasma step of this same plasma recipe A, respectively. Each of thesespectra744,752, and760 are characterized by a number of peaks748,756, and764, respectively, of varying intensities at various wavelengths. A comparison of thespectra744,752, and760 reveals that their associated patterns differ, including without limitation as follows: 1) at about the 425 nanometer wavelength region, peak748ain thespectra744 of FIG. 17A has an intensity of about 3,300, peak756ain thespectra752 of FIG. 17B has an intensity of about 3,000, andpeak 764ain thespectra760 of FIG. 17C has an intensity of about 2,100; 2) at about the 475 nanometer wavelength region, peak748bin thespectra744 of FIG. 17A has an intensity of about 3,200, peak756ain thespectra752 of FIG. 17B has an intensity of about 3,900, and there is no peak in thespectra760 of FIG. 17C, but the corresponding intensity (noise) is about 500; 3) at about the 525 nanometer wavelength region, peak748cin thespectra744 of FIG. 17A has an intensity in excess of 4,000,peak756cin thespectra752 of FIG. 17B has an intensity of about 3,400, and peak764cin thespectra760 of FIG. 17C has an intensity of about 2,750; 4) at about the 587 nanometer wavelength region, there is no peak in thespectra744 of FIG. 17A, but the intensity is about 500 (noise), there is no peak in thespectra752 of FIG. 1713, but the intensity is about 490 (noise), and peak764din thespectra760 of FIG. 17C has an intensity of about 3,000. These distinctions between thespectra744,752, and760 show that it is possible to distinguish between the various steps of a plasma recipe, as well to evaluate the progression of a plasma recipe, through an evaluation of the spectral pattern of the plasma in thechamber36 during the plasma recipe.

Exemplary Plasma Cleaning Operation Spectra with No Prior Wet Clean—FIGS.18A-C

Representative spectra are presented in FIGS. 18A-C to illustrate how the optical emissions of the plasma within theprocessing chamber36 changes over time during a plasma clean conducted without first doing a wet clean of thechamber36. FIG. 18A presents aspectra770 of an exemplary plasma when theprocessing chamber36 is in a dirty chamber condition and while plasma is present in theprocess chamber36 without any product therein. FIG. 18B presents aspectra774 of this same exemplary plasma at an intermediate time of the plasma clean in which the dirty chamber condition has begun to be addressed by the plasma clean. Finally, FIG. 18C presents aspectra778 of this same exemplary plasma at the end of the plasma clean at which time the interior of theprocessing chamber36 is deemed to be in condition to return to commercial production (e.g., to etch integrated circuit designs on a production wafer18). Thisspectra778 may be selected by the operator of the facility implementing thewafer production system2 as being indicative of thechamber36 being in proper condition for resumption of production. However, it should be appreciated that there is not necessarily a “bright line” as to when achamber474 is in condition to return to production. Therefore, the selection of thespectra778 as being indicative of a “clean chamber condition” may be somewhat arbitrary.

Each of thespectra770,774, and778 are characterized by a number of peaks772,776, and780, respectively, of varying intensities at various wavelengths. A comparison of thespectra770,774, and778 reveals that their associated patterns are in fact different, including without limitation as follows: 1) at about the 625 nanometer wavelength region, peak772ein thespectra770 of FIG. 18A has an intensity of about 500, peak776ein thespectra774 of FIG. 18B has an intensity of about 300, and there is no substantial peak in thespectra778 of FIG. 18C; 2) at about the 675 nanometer wavelength region, peak772fin thespectra770 of FIG. 18A has an intensity of about 4,000,peak776fin thespectra774 of FIG. 18B has an intensity of about 1,000, and there is no substantial peak in thespectra778 of FIG. 18C; and 3) at about the 685 nanometer wavelength region, peak772gin thespectra770 of FIG. 18A has an intensity of about 3,400, peak776gin thespectra774 of FIG. 18B has an intensity of about 2,200, and peak780gin thespectra778 of FIG. 18C has an intensity of about 700. These distinctions between thespectra770,774, and778 show the progression of a plasma clean is evident in the spectral pattern of the plasma in thechamber36 during the plasma clean.

More than one entry of a plasma cleaning may be required in thenormal spectra subdirectory288 depending upon a variety of factors. For instance, the spectral data of a plasma cleaning which is run after thechamber36 has been running a first type of plasma recipe may look different than a plasma cleaning which is run after thechamber36 has been running a second type of plasma recipe which is different from the first type of plasma recipe.

Exemplary Plasma Cleaning Operation Spectra Conducted After Wet Clean—FIGS.19A-C

Representative spectra of one plasma cleaning operation of thechamber36 after it has been wet cleaned are illustrated in FIGS. 19A-C. FIG. 19A presents aspectra1328 of the exemplary plasma in the processing chamber at the start of such a plasma cleaning of thechamber36, while FIG. 19B presents aspectra1336 of the exemplary plasma at an intermediate point in such a plasma cleaning of thechamber36, and while FIG. 19C presents aspectra1344 of the exemplary plasma at the end of such a plasma cleaning of thechamber36. Each of thespectra1328,1336, and1344 are characterized by a number of peaks1332,1340, and1348, respectively, of varying intensities at various wavelengths. A comparison of thespectra1328,1336, and1344 reveals that their respective patterns are different, including without limitation as follows: 1) at about the 625 nanometer wavelength region, peak1332ein thespectra1328 of FIG. 19A has an intensity of about 600, peak1340ein thespectra1336 of FIG. 19B has intensity of about 500, and peak1348ein thespectra1344 of FIG. 19C has an intensity of about 450; 2) at about the 668 nanometer wavelength region, peak1332fin thespectra1328 of FIG. 19A has an intensity in excess of 4,000, peak1340fin thespectra1336 of FIG. 50B has intensity of about 1,000, and peak1348fin thespectra1344 of FIG. 19C has an intensity of about 400; and 3) at about the 685 nanometer wavelength region, peak1332gin thespectra1328 of FIG. 19A has an intensity in excess of 3,400, peak1340gin thespectra1336 of FIG. 19B has intensity of about 2,300, and peak1348gin thespectra1344 of FIG. 19C has an intensity of about 1,400. These distinctions between thespectra1328,1336, and1344 show that the progression of a plasma cleaning operation is evident in the spectral pattern of the plasma in thechamber36 during the plasma clean.

More than one entry of a plasma clean may be required in thenormal spectra subdirectory288 depending upon a variety of factors. For instance, the spectral data of a plasma clean run on achamber36 after a wet clean may look different than a plasma clean that is run on anew chamber36 which has not been wet cleaned. Moreover, the spectral data of a plasma clean which is run after thechamber36 has been running a first type of plasma recipe may look different than a plasma clean which is run after thechamber36 has been running a second type of plasma recipe which is different from the first type of plasma recipe.

Exemplary Conditioning Wafer Operation Spectra—FIGS.20A-C

Representative spectra of a conditioning wafer operation are illustrated in FIGS. 20A-C. FIG. 20A presents aspectra1288 of an exemplary plasma in theprocessing chamber36 at the start of a conditioning wafer operation, FIG. 20B presents aspectra1292 of an exemplary plasma at an intermediate point in the conditioning wafer operation, and FIG. 20C presents aspectra1296 of an exemplary plasma at the end of the conditioning wafer operation. Each of thespectra1288,1292 and1296 are characterized by a number of peaks1290,1294, and1298, respectively, of varying intensities at various wavelengths. A comparison of thespectra1288,1292, and1296 reveals that there are certain differences in their respective patterns, including without limitation as follows: 1) at about the 440 nanometer wavelength region, peak1290ain thespectra1288 of FIG. 20A has an intensity of about 3,550, peak1294ain thespectra1292 of FIG. 20B has an intensity of about 3,750, and peak1298ain thespectra1292 of FIG. 20C has an intensity of about 4,000; 2) at about the 525 nanometer wavelength region, peak1290bin thespectra1288 of FIG. 20A has an intensity of about 2,800, peak1294bin thespectra1292 of FIG. 20B has an intensity of about 2,900, and peak1298bin thespectra1296 of FIG. 20C has an intensity of about 2,800; 3) at about the 595 nanometer wavelength region, peak1290din thespectra1288 of FIG. 20A has an intensity of about 2,100, peak1294din thespectra1292 of FIG. 20B has an intensity of about 2,150, and peak1298din thespectra1296 of FIG. 20C has an intensity of about 2,125; 4) at about the 675 nanometer wavelength region, peak1290ein thespectra1288 of FIG. 20A has an intensity of about 600, peak1294ein thespectra1292 of FIG. 20B has an intensity of about 250, and there is no peak in thespectra1296 of FIG. 20C; and 5) at about the 685 nanometer wavelength region, peak1290fin thespectra1288 of FIG. 20A has an intensity of about 1,450, peak1294fin thespectra1292 of FIG. 20B has an intensity of about 600, and there is no peak in thespectra1296 of FIG.20C. These distinctions between thespectra1288,1292, and1296 show that the progression of a conditioning wafer operation is evident in the spectral pattern of the plasma in thechamber36 during the operation.

More than one entry of a conditioning wafer operation may be required in thenormal spectra subdirectory288 depending upon a variety of factors. For instance, the spectral data of a conditioning wafer operation which is run after thechamber36 has only been plasma cleaned may look different than the spectral data of a conditioning wafer operation which is run in thechamber36 after it has been both plasma cleaned, wet cleaned, and then again plasma cleaned. Moreover, the spectral data of a conditioning wafer operation run after thechamber36 has been running a first type of plasma recipe may look different than a conditioning wafer operation which is run after thechamber36 has been running a second type of plasma recipe which is different from the first type of plasma recipe.

Plasma Health Module252—FIGS.21-25

The currentplasma process module250 of FIGS. 7 and 32 is available for monitoring the health of any plasma process which is conducted within theprocessing chamber36 first through a comparison of at least a portion of its spectral data with at least a portion of the spectral data stored in the normal spectra subdirectory288 (FIG.9). Plasma recipes (whether run onproduction wafers18 or qualification wafers18), plasma cleans (with or without wet cleans), and conditioning wafer operations, as well as the health of any other plasma process, may each be evaluated through theplasma health module252. How theplasma health module252 deals with the presence of having different categories of plasma processes stored in thenormal spectra subdirectory288, as well as in theabnormal spectra subdirectory292 and theunknown spectra subdirectory296 which are also used in the plasma health evaluation of a current plasma process being run in thechamber36, is really a matter of preference. Some ways of dealing with the existence of multiple categories of plasma processes may affect the speed of the evaluation by theplasma health module252 more than others. For instance, theplasma health module252 may limit its comparison of the current plasma process to the same category or genus of plasma processes stored in thenormal spectra directory288 andabnormal spectra directory292. Appropriate “identifying information” may be input into the plasmaprocess genus field322h(FIG. 12A) associated with each plasma process stored in thenormal spectra subdirectory288 and the plasma process genus field333e(FIG. 12B) associated with each plasma process (or portion thereof) stored in theabnormal spectra subdirectory292. Moreover, the current plasma process to be conducted in thechamber36 may be identified to theplasma health module252 in some manner. This may be accomplished through thestartup module202 of FIG. 15 (e.g., through including appropriate process category or genus identifying information insteps230,234,238, and242 of thestartup subroutine203, which is passed ontostep236 of thesubroutine203, and which may be then passed on to the plasma health module252). Reducing the number of entries in thenormal spectra subdirectory288 andabnormal spectra subdirectory292 which are compared with the current plasma process may and typically will increase the speed of evaluation of the health of the current plasma process by theplasma health module252. However, there may be advantages to not imposing a plasma process category or genus match criterion when selecting data from thenormal spectra subdirectory288 and/orabnormal spectra subdirectory292 which is to be available for comparison with the current plasma process.

Plasma health is also preferably evaluated by comparing optical emissions from the current plasma process in thechamber36 with theplasma spectra directory284 over at least those wavelengths within the Preferred Optical Bandwidth based upon the Preferred Data Resolution, and using the Preferred Analytical Wavelength Resolution. However, in some cases some subset of the optical emissions data of the plasma in thechamber36 may be used to monitor the plasma health. One such circumstance is when processing speed is or potentially is an issue. There are a number of ways of selecting the amount of optical emissions data to monitor the health of a plasma process. The data within theabnormal spectra subdirectory292 of FIG. 9 may be used to generate the subset of data which may be reviewed for purposes of monitoring the plasma health. For instance, the plasma health evaluation may be conducted over optical emissions segments which include those wavelengths which are indicative of errors which occurred in processes previously conducted within thechamber36. One alternative is to define an optical emissions segment ±25 nanometers on each side of each wavelength of a spectra in theabnormal spectra subdirectory292 which is indicative of an error from a previous plasma process. For instance, if errors from previous runs are reflected at the 325, 425, and 575 wavelengths, the plasma health may be evaluated by looking at each of the 300-350, 400-450, and 550-600 nanometer region. A smaller optical emissions segment for monitoring plasma health may also be selected by defining a range which includes each of those wavelengths which are indicative of errors from theabnormal spectra subdirectory292. For instance, if errors from previous runs are reflected at the 325, 425, and 575 wavelengths, the plasma health may be evaluated by looking at the wavelength region from about 325 nanometers to about 575 nanometers. It may be desirable to include a “buffer” on each of the endpoints of this range as well (e.g., extend by about 25 nanometers on each end of the range). The above may be further limited by limiting the plasma health evaluation to those optical emissions segments which include only errors from the same type of plasma process which is to be run in the chamber36 (e.g., same plasma recipe). Finally, information on endpoint of the plasma process or discrete portion thereof may be used to define the wavelengths to be evaluated in relation to plasma health. As will be discussed in more detail below in relation to theendpoint detection module1200 and FIG. 52, endpoint may be called based upon a change at one or more specific wavelengths. Plasma health may be evaluated by looking at a ±25 nanometer region around each wavelength which is used to call endpoint. Notwithstanding the foregoing, plasma health should be evaluated over at least a 50 nanometer wavelength in which data has been collected at the Preferred Data Resolution, and again is preferably undertaken using the Preferred Optical Bandwidth.

Plasma Health Subroutine253—FIG.21

One embodiment of a subroutine is illustrated in FIG. 21 which may be used by theplasma health module252 to evaluate whether a current plasma process is proceeding in accordance with at least one plasma process stored in thenormal spectra subdirectory288 of FIG. 9 (e.g., indicative of a “healthy” plasma). Summarily, spectral data is taken during and more preferably throughout the entirety of the execution of the current plasma process which is being run within theprocessing chamber36. Consideration should be given to the first part of a plasma process being somewhat unstable. Spectral data from the current plasma process is first compared against thenormal spectra subdirectory288 to determine if the current plasma process “matches” any plasma process stored within thenormal spectra subdirectory288. As long as the current plasma process “matches” at least one plasma process stored in thenormal spectra subdirectory288, the current plasma process is characterized as being “normal” or “healthy” and theplasma health subroutine253 will continue to limit its search for “matching” spectra to thenormal spectra subdirectory288. However, oftentimes there is an error or aberration during a plasma process which may have some type of adverse effect on the desired end result of the plasma process, and this should be identifiable from the spectra of the plasma in thechamber36.

During an error or aberration in a plasma process currently being run, the spectra of the plasma in thechamber36 should no longer “match” any plasma process stored in thenormal spectra subdirectory288. Theplasma health subroutine253 will then discontinue its search of thenormal spectra subdirectory288 for evaluating the current plasma process and start comparing the current plasma process with theabnormal spectra subdirectory292 of FIG.9. Errors or aberrations in plasma processes which have been encountered before by theplasma health subroutine253 on thissame chamber36, and which have had their corresponding cause or causes identified, are recorded in theabnormal spectra subdirectory292. Actions which may be initiated if spectral data of the current plasma process “matches” at least one spectra in theabnormal spectra subdirectory292 range from issuing an appropriate alert to addressing one or more process control features of thewafer production system2 as discussed above in relation to theprocess alert subroutine432 of FIG.14.

All of the data in thenormal spectra subdirectory288 and theabnormal spectra subdirectory292 is obtained from theprocessing chamber36 on which theplasma health module252 is being used to evaluate any plasma process currently being run or which was run in this verysame chamber36. Building of the library of information for thenormal spectra subdirectory288 andabnormal spectra subdirectory292 takes time to allow theplasma health subroutine253 to “learn” from thechamber36 and the plasma processes being run therein. Circumstances will therefore likely arise where the spectral data of the plasma from the current plasma process cannot be found in either thenormal spectra subdirectory288 or theabnormal spectra subdirectory296 by theplasma health subroutine253. Information of this type is stored for thesubroutine253 in theunknown spectra subdirectory296 of FIG.9. It will likely remain an “unknown condition” until the spectral data can be properly analyzed and the “cause” identified, at which time the relevant spectral data may be transferred to either thenormal spectra subdirectory288 or theabnormal spectra subdirectory292 to upgrade the plasma health subroutine's253 knowledge of theprocessing chamber36 and its associated plasma processes.

Specifics of theplasma health subroutine253 will now be addressed in the case where the plasma process is a plasma recipe being run onwafers18 in thechamber36. An exemplary general procedure for running a plasma recipe onwafers18 is as follows. First, acassette6 having a plurality ofproduction wafers18 therein, as well as possibly one or more qualification wafers, is transferred into one of the load lock chambers28 (FIG.1). Thewafer handling assembly44 will retrieve one of thewafers18 from thecassette6 and transport the same into theprocessing chamber36. At this time, the plasma in thechamber36 is off. Thechamber36 is sealed and the plasma is ignited to run a plasma recipe on thewafer18. Typical practice is to run the same plasma recipe on theentire cassette6. After the completion of the plasma recipe on thefirst wafer18, the plasma is turned off, thechamber36 is opened, and thewafer handling assembly44 retrieves thewafer18 from thechamber36 and provides the same back to its corresponding slot in thecassette6. Once all of thewafers18 of thecassette6 have been processed in this manner (usually 1-3 qualification wafers are used for acassette6 having 24wafers18 and may be included anywhere within the cassette6), thecassette6 may be removed from the load lock chamber28 and replaced with anothercassette6 ofwafers18. Thequalification wafers18 from the plasma processedcassette6 may then be tested (destructively or non-destructively), while semiconductor devices may be formed from theproduction wafers18.

Plasma processes, whether are run on qualification orproduction wafer18, in theprocessing chamber36 are evaluated by theplasma health module252. Usually no more than a 1 minute time lapse exists between the time oneproduction wafer18 is removed from theprocessing chamber36 and the time the plasma recipe is initiated on thenext production wafer18 which is loaded into thechamber36. Theplasma health module252 is able to complete its evaluation of a plasma recipe which was run on aproduction wafer18 before the plasma recipe is initiated on thenext production wafer18 since theplasma health module252 effectively relies on pure pattern recognition techniques, and not chemical analysis or chemical species identification techniques. Moreover and as will be discussed in more detail below, theplasma health module252 is able to not only determine the identify of the plasma process, but to determine that the plasma process is being run on aqualification wafer18 versus aproduction wafer18.

Data which relates to the current plasma process being run on product in theprocessing chamber36 at the current time t_cis obtained for evaluation by theplasma health subroutine253 of FIG.21 through execution ofstep254. Althoughstep254 is referenced in FIG. 21 in relation to merely a “spectra” or optical emissions data, as noted above, other types of data may be taken at/associated with this time as well (e.g., the time into the plasma recipe at which the associated spectra was obtained from the chamber36 ). A comparison is then made atstep258 of theplasma health subroutine253 between the spectra of the current plasma process obtained at step254 (current plasma process) and the relevant spectra from the normal spectra subdirectory288 (stored plasma process). For facilitating the initial understanding of how the health of a plasma process may be evaluated by theprocess health module252, thenormal spectra directory288 will hereafter be described as having only a single plasma process stored therein (“Recipe A”). How the health of a current plasma process may be handled by theplasma health module252 in the situation where multiple plasma processes are stored in thenormal spectra subdirectory288 is addressed in the discussion of the plasma health/process recognition subroutines790,852, and924 presented below in relation to FIGS. 22-24.

Step258 of theplasma health subroutine253 calls thepattern recognition module370 of FIG. 13 to undertake a comparative analysis between the spectra of the plasma in thechamber36 at the current time t_cand the relevant spectra of Recipe A from thenormal spectra subdirectory288. This is affected bystep258 of thesubroutine253 setting the Target Directory used by thepattern recognition module370 to thenormal spectra subdirectory288. Only thenormal spectra subdirectory288 is then searched by thepattern recognition module370 at this time, through execution ofstep258 of theplasma health subroutine253, to determine if there is a “match” between the current spectra at the current time t_cand Recipe A as stored in thenormal spectra subdirectory288. Which spectra of Recipe A is actually compared with the current spectra in this instance is addressed below after the discussion of theloop190 of thesubroutine253 is completed.

Thepattern recognition module370 of FIG. 13 returns control of the plasma monitoring operation back to theplasma health subroutine253 of FIG. 21 after thepattern recognition module370 has determined whether there is a “match” between the spectra at the current time t_c(fromstep254 of the plasma health subroutine253) and the relevant spectra of Recipe A from thenormal spectra subdirectory288 in the subject example. The result (“match” or “no match”) of the analysis by thepattern recognition module370 is actually provided to step260 of theplasma health subroutine253 of FIG.21. If the current spectra at the current time t_cwas a “match” with the relevant spectra of Recipe A, the evaluation by theplasma health subroutine253 will continue in relation to thenormal subdirectory288. In this regard, theplasma health subroutine253 inquires atstep261 as to whether the current plasma process being conducted in theprocessing chamber36 has been terminated, or more accurately if there are any more spectra from the subject current plasma process being run in thechamber36 to be evaluated by thesubroutine253. Other information on the subject current plasma process may be provided through execution ofstep194 of theplasma health subroutine253 which calls other sub-modules of the currentplasma process module250 as will be discussed in more detail below (e.g., to access a chamber condition evaluation function, to access an endpoint determination function).

Additional optical emissions data on the current plasma process for evaluation by theplasma health subroutine253 is made available through execution ofstep278. Atstep278 the “clock” of thesubroutine253 is effectively reset. Step278 of thesubroutine253 more specifically provides for adjustment of the “clock” by a predetermined increment “n” to thereby increase the current time t_cby an increment of “n.” The magnitude of “n” defines that portion of the collected data which will be analyzed. All of the data may be analyzed, or only a portion thereof (e.g., only every other “piece” of optical emissions data may actually be analyzed). Hereafter, this concept will be referred to as the Analytical Time Resolution. In one embodiment, the Analytical Time Resolution in relation to plasma health is at least at every 1 second, and more preferably at least at every 300 milliseconds. Theplasma health subroutine253 then returns to step254 where the next spectra of the plasma in theprocessing chamber36 from the execution of the current plasma process is obtained for thesubroutine253 at the new current time t_csuch that the same may be evaluated in accordance with the foregoing.

There are a number of ways to define what is the “relevant spectra” of a plasma process stored in thenormal spectra subdirectory288 for comparison with the current spectra of the plasma in thechamber36 at the current time t_cthrough theplasma health subroutine253. Relevance may be time dependent and will hereafter be referred to as a “time dependency requirement.” In the subject example where the only plasma process stored in thenormal spectra subdirectory288 is Recipe A, the relevant spectra of Recipe A to be compared with the spectra of the plasma in thechamber36 at the current time t_cwould be limited to that spectra for Recipe A which is associated with this same current time t_cif a time dependency requirement was used. That is, the spectra of the plasma in thechamber36 during the current plasma process at time t₁would be compared with Recipe A's spectra which is associated with the same time t₁through execution ofstep258 of theplasma health subroutine253, the spectra of the plasma in thechamber36 during the current plasma process at time t₂would be compared with the Recipe A's spectra which is associated with the same time t₂through execution ofstep258 of thesubroutine253, and so forth. This time dependency requirement for determining what is the “relevant spectra” may be used regardless of which subdirectory of theplasma spectra directory284 is being “searched” by thepattern recognition module370.

In theory, a time dependency requirement is an acceptable way to evaluate whether a current plasma process is proceeding in accordance with any one or more plasma processes stored in thenormal spectra subdirectory288. From a practical standpoint this is not necessarily the case. Variations throughout thewafers18 in a givenwafer cassette6 on which the same plasma recipe is typically run may affect the amount of time required to complete one or more of the plasma steps of the current plasma recipe being run in thechamber36. For instance, the thickness of a certain layer to be etched away by a certain step of the plasma recipe may vary fromwafer18 towafer18 within an acceptable tolerance. Conditions within thechamber36 may also have an effect on the amount of time required to reach the endpoint of one or more plasma steps of a given plasma recipe or any other plasma process for that matter. For instance, as the interior of thechamber36 ages by the formation of deposits within the interior, by the etching away of materials from the interior, or both, the performance of thechamber36 may change. Changing the performance of thechamber36 may change the amount of time required to reach the endpoint of one or more steps of a given plasma recipe. Other factors may affect timing issues associated with other types of plasma processes run in thechamber36. Failing to account for these types of factors will result in theplasma health subroutine253 issuing false alarms, or more specifically an indication that a current plasma process does not conform with at least one plasma process stored in thenormal spectra subdirectory288 when such is not the case.

Alternatives to the time dependency requirements exist for determining what is the “relevant spectra” of a given plasma process stored in thenormal spectra subdirectory288 for comparison with the spectra of the plasma in thechamber36 at the current time t_cfrom the current plasma process. “Relevant” in the context of theplasma health subroutine253, and in fact for each sub-module of the currentplasma process module250, may simply be whether the current plasma process being run in thechamber36 is progressing in a manner consistent with at least one of the plasma processes stored in thenormal spectra subdirectory288, although not necessarily at the same speed and therefore not being time dependent. Hereafter this will be referred to as a “progression dependency requirement” and is exemplified by the following. The first spectra obtained for theplasma health subroutine253 at the current time t₁is compared with one or more spectra of Recipe A in thenormal spectra subdirectory288 in the subject example. The spectra of Recipe A which matched the spectra of the plasma in thechamber36 at the current time t₁, and which has the earliest time associated therewith, is identified as the current status spectra of Recipe A. This accounts for the possible, although likely improbable, situation where a spectra at time t₁in a given plasma process is substantially the same as a spectra at, for instance, time t₁₀₀in this same plasma process. The next spectra obtained for theplasma health subroutine253, or the spectra at time t₂in the example, is first compared with this same current status spectra of Recipe A in thenormal spectra subdirectory288. If the current spectra at the current time t₂still matches this current status spectra of Recipe A, the current status spectra of Recipe A remains unchanged. However, if the spectra of the plasma in thechamber36 from the current plasma recipe being run on product in thechamber36 at the current time t₂does not match the current status spectra of Recipe A, thepattern recognition module370 will look to see if this current spectra matches the spectra of Recipe A which follows (in time) the current status spectra in Recipe A. A “match” between the current spectra of the plasma in theprocessing chamber36 at the current time t_cand the spectra of Recipe A which follows the current status spectra means that the current plasma process is progressing accordingly, and this later in time “matching spectra” in Recipe A is now the current status spectra of Recipe A. The foregoing logic will continue to repeated so long as the current plasma process is progressing in accordance with Recipe A of thenormal spectra subdirectory288. This progression dependency requirement may be used regardless of which subdirectory of theplasma spectra directory284 is being searched by thepattern recognition module370. Moreover, the logic may be that the current spectra is first checked against the next spectra following (in time) the current status spectra, and then back against the current status spectra only if there is not match for purposes of determining if the process is progressing (e.g., assume progression at the same rate and only look back if necessary).

Theloop190 defined bysteps254,258,260,261, and278 of theplasma health subroutine253 will continue to be re-executed until one of two events occurs. One event which will cause theplasma health subroutine253 to exit theloop190 is when all of the spectral data on the current plasma process has been evaluated by thesubroutine253 in accordance with the foregoing. That is, the entirety of the current plasma process run in theprocessing chamber36 proceeded in accordance with at least one of the plasma processes recorded in the normal spectra subdirectory288 (“at least one” referring to the fact that more than one entry of a given plasma process may be included in the normal spectra subdirectory288), or in the subject example Recipe A. Control of the plasma monitoring operations is then transferred from theplasma health subroutine253 back to, for instance, thestartup module202 of FIG.15 through execution ofstep279 of theplasma health subroutine253. It should be noted that the results of “normal” runs may be recorded in a “normal run” log file. Data such as that presented in FIG. 12A may be included in this “normal run” log file and will provide a historical record of the subject plasma process. If data storage space is an issue, the spectral data may be omitted from the historical record although it is desirable to retain this data. Moreover, this historical data need not be stored for access by the currentplasma process module250. For instance, the historical data may be stored on a network associated with thewafer production system2 or any other data storage area.

Theplasma health subroutine253 may also exit the loop190 (the evaluation of the current spectra at the current time t_cin relation to thenormal spectra subdirectory288 of FIG. 9) when this current spectra is no longer a “match” with any plasma process stored in thenormal spectra subdirectory288. This would be the case when at some point in time the current plasma recipe being run on product within theprocessing chamber36 was not a “match” with Recipe A of thenormal spectra subdirectory288 in the subject example. The results of thepattern recognition module370 of FIG. 13 provided back to step260 of theplasma health subroutine253 in this case would cause thesubroutine253 to proceed fromstep260 to step266.

Step266 of theplasma health subroutine253 calls thepattern recognition module370 of FIG. 13 to undertake a comparative analysis between the spectra of the plasma in thechamber36 at the current time t_cand the relevant spectra of theabnormal spectra subdirectory292. This is accomplished bystep266 of thesubroutine253 setting the Target Directory used by thepattern recognition module370 to theabnormal spectra subdirectory292. Only theabnormal spectra subdirectory292 is then searched by thepattern recognition module370 through execution ofstep266 to determine if there is a “match” between the current spectra of the plasma in thechamber36 at the current time t_cand the relevant spectra stored in theabnormal spectra subdirectory292.

A number of options exist in relation to which spectra of theabnormal spectra subdirectory292 of FIG. 9 are actually compared with the current spectra of the plasma in theprocessing chamber36 at the current time t_cusing the above-described analysis provided by thepattern recognition module370. Each spectra stored within theabnormal spectra subdirectory292 may and preferably does have a time associated therewith, which is the time into the plasma process in which the spectra was obtained from within the chamber36 (i.e., its corresponding t_c). The search of theabnormal spectra subdirectory292 for “matches” by thepattern recognition module370 may be limited to those spectra which were recorded at the same current time t_cor within a predetermined amount of time on each side of the subject current time t_c(e.g., ±“x” seconds of the subject current time t_c). For example, if the current spectra was obtained 20 seconds into a plasma process being run in theprocessing chamber36, the point-by-point analysis embodied by step386 of thepattern recognition subroutine374 of FIG. 13 may be performed in relation to only those spectra within theabnormal spectra subdirectory292 which were also recorded at the same 20 second time period or within ±10 seconds (or any other desired amount) of this time period.

Another subset of theabnormal spectra subdirectory292 which may be used as a refining search criteria is the plasma process category/genus, or even the plasma process type or species within a plasma process category/genus having multiple types/species of plasma processes (e.g., a specific type of plasma recipe). That is, only those spectra in theabnormal spectra subdirectory292 which are associated with a plasma process which is at least possibly the same as that currently being run on product within theprocessing chamber36 will be analyzed by thepattern recognition module370 using a plasma process criterion. “Potential” matching processes is used in this situation because at the time of the error in the current plasma process, theplasma health module252 may not have narrowed down the identification of the current plasma process to a single plasma process within thenormal spectra subdirectory288. How theplasma health module252 may identify a current plasma process being run on product in theprocessing chamber36 is addressed below in relation to the plasma health/process recognition subroutines790,852, and924 of FIGS. 22-24.

The plasma step of a plasma process may also be used as a refining search criterion for which spectra of theabnormal spectra subdirectory292 are analyzed by thepattern recognition module370. That is, only those spectra in theabnormal spectra subdirectory292 which are associated with a plasma step of a multiple step plasma process which is still possibly a match with a current plasma step being run within theprocessing chamber36 will be analyzed by thepattern recognition module370 in this case. “Potential” is used in relation to matching plasma steps of plasma processes in this situation because at the time of the error in the current plasma process, theplasma health module252 may not have narrowed down the identification of the current plasma step and process to a single plasma step of a single plasma process within thenormal spectra subdirectory288. How theplasma health module252 may identify a current plasma step of a current plasma process being run in theprocessing chamber36 is through the plasma health/processstep recognition subroutine972 which will be discussed below in relation to FIG.25.

Any combination of the foregoing may be used as initial search criteria to initially refine the search of theabnormal spectra subdirectory292. Finally, no refining search criteria need be used. That is, the search of theabnormal spectra subdirectory292 for “matches” may compare the current spectra at the current time t_cwith each spectra within theabnormal spectra subdirectory292, thereby removing both the time element and plasma process category/plasma process type within a given plasma process category/plasma step element as required “initial match” criteria.

Thepattern recognition module370 returns control of the plasma monitoring operation back to theplasma health subroutine253 of FIG. 21 after thepattern recognition module370 has determined whether there is a “match” between the spectra of the plasma in thechamber36 at the current time t_c(fromstep254 of the plasma health subroutine253) and the relevant spectra from theabnormal spectra subdirectory292. The result (“match” or “no match”) of the analysis by thepattern recognition module370 is provided to step276 of theplasma health subroutine253 of FIG.21. If the current spectra of the plasma in theprocessing chamber36 at the current time t_cwas a “match” with at least one spectra in theabnormal spectra subdirectory292, theplasma health subroutine253 effectively takes two actions and these actions may be undertaken in any order, including simultaneously. One of these actions is that theplasma health subroutine253 will proceed to step274 which calls theprocess alert module428 which was discussed above in relation to FIG.14. Generally, alerts may be issued as to the identification of the abnormal condition, control of thewafer production system2 may be addressed, or both through theprocess alert module428. Another action which is taken by theplasma health subroutine253 in this type of case is to record data of the remainder of the plasma process in an “abnormal run” log file for historical purposes.

At a minimum, spectral data is recorded in an “abnormal run” log file through execution ofstep264 for the current time t_c(the first spectra which did not “match” any relevant plasma process stored in thesubdirectory288, but which did “match” at least one entry in the abnormal spectra subdirectory292). Proceeding fromstep264 to step265 of theplasma health subroutine253 of FIG. 21, a determination is made as to the status of the current plasma process. Any termination of the plasma process, or more accurately recordation of data on the remainder of the process, will cause thesubroutine253 to proceed to step267 where control of the plasma monitoring operations may be returned to, for instance, thestartup module202 of FIG.15. Continuation of the plasma process after the error is identified will cause thesubroutine253 to proceed fromstep265 to step268 where the current time t_cis increased by a factor of “n” such that another spectra of the plasma in thechamber36 can be obtained for thesubroutine253 at this new current time t_cthroughstep272 for recordation in the “abnormal run” log file. The magnitude of “n” may be the Preferred Analytical Time Resolution.Steps264,265,268, and272 will continue to be repeated in the described manner to continue to record data in the “abnormal run” log file until the current plasma process is terminated, at which time thesubroutine253 will exit atstep267 as described.

Circumstances will be encountered where the current spectra of the plasma in theprocessing chamber36 at the current time t_cwill not “match” any plasma process stored in thenormal spectra subdirectory288, and further where this current spectra will not “match” any relevant spectra stored in theabnormal spectra subdirectory292. This is referred to herein as an “unknown condition.” Theplasma health subroutine253 of FIG. 21 handles this type of situation by exitingstep276 to where thesubroutine253 effectively takes two actions, and these actions may be undertaken in any order and including simultaneously. One of these actions is that theplasma health subroutine253 will execute step256 which calls theprocess alert module428 discussed above in relation to FIG.14. Generally, alerts may be issued as to the existence of the unknown condition, control of thewafer production system2 may be addressed, or both through theprocess alert module428. Another “action” which is taken by theplasma health subroutine253 in this type of case is to record data of the remainder of the plasma process in theunknown spectra subdirectory296.

At a minimum, spectral data is recorded in theunknown spectra subdirectory296 through execution ofstep270 for the current time t_c(the first spectra which did not “match” any relevant plasma process stored in thenormal spectra subdirectory288, and which also did not “match” any relevant entry in the abnormal spectra subdirectory292). Proceeding fromstep270 to step283 of theplasma health subroutine253, a determination is made if the current plasma process has been terminated (e.g., is the plasma “off” in the chamber36 ). Any termination of the plasma process will cause thesubroutine253 to proceed to step281 where control of the plasma monitoring operations may be returned to, for instance, thestartup module202 of FIG.15. Continuation of the plasma process after the unknown condition is encountered will cause thesubroutine253 to proceed fromstep283 to step280 where the current time t_cis increased by a factor of “n” (e.g., Preferred Data Collection Time Resolution) such that another spectra of the plasma in thechamber36 can be obtained for thesubroutine253 at this new current time t_cthroughstep282 for recordation in theunknown spectra subdirectory296.Steps270,283,280, and282 will continue to be repeated in the described manner until the current plasma process is terminated or until data on the remainder of the process has been recorded in thesubdirectory296, at which time thesubroutine253 will exit atstep281 as described.

Plasma process runs which are recorded in theunknown spectra subdirectory296 are typically subsequently analayzed to attempt to identify the cause of the unknown condition as noted above. If the unknown condition actually turns out to be a new plasma process, the data recorded in theunknown spectra subdirectory296 may be transferred to thenormal spectra subdirectory288. New spectral patterns from this new plasma process will then be available to assess further runnings of plasma processes on thissame processing chamber36. If the unknown condition turns out to be an error with an associated cause, some or all of the data from the subject run recorded in theunknown spectra subdirectory296 may be transferred to theabnormal spectra subdirectory292. At least one new spectral pattern which is representative of the newly identified error condition will then be available to assess further runnings of plasma processes on thissame processing chamber36 through theplasma health module252.

Plasma Health/Process Recognition subroutine790—FIG.22

Another embodiment of a subroutine which may be used by theplasma health module252 is presented in FIG.22. Not only is the health or condition of the plasma assessed by thesubroutine790 in FIG. 22, but the particular plasma process which is being run in thechamber36 is also identified. That is, thesubroutine790 is able to determine the identify of the plasma process (e.g., to distinguish between different types of plasma recipes, to distinguish between the same plasma recipe run on aproduction wafer18 and aqualification wafer18, etc). Consequently, thesubroutine790 is characterized as a plasma health/process recognition subroutine790. The plasma health/process recognition subroutine790 also presents one way in which a current plasma process being run in thesubject chamber36 may be evaluated against multiple plasma processes stored in thenormal spectra subdirectory288 of FIG.9. These very same principles may be implemented in theplasma health subroutine253 of FIG.21.

A number of prerequisites are addressed before the plasma health/process recognition subroutine790 actually initiates its analysis of the current plasma process being run in theprocessing chamber36. The order in which these steps are executed is not important to the present invention. Initialization of the plasma health/process recognition subroutine790 includes setting the Target Directory associated with thepattern recognition module370 of FIG. 13 to thenormal spectra directory288 of FIG. 9 atstep796 of thesubroutine790. Generally, thepattern recognition module370 is used by thesubroutine790 to compare the pattern of a “run spectra” (i.e., a spectra of the plasma from theprocessing chamber36 during a plasma process being run in the processing chamber36) with the relevant spectra of the plasma processes stored in thenormal spectra subdirectory288.

Preparation for the analysis of the current plasma process being run in thechamber36 by the plasma health/process recognition subroutine790 also requires execution ofstep816 which “calls up” or “flags” the first plasma process in thenormal spectra subdirectory288 which is to be compared with the current plasma process by thesubroutine790. The logic of thesubroutine790 is to compare the current plasma process being run in theprocessing chamber36 with only one plasma process stored in thenormal spectra subdirectory288 at a time. That is, thesubroutine790 will first compare the current plasma process with Process A in thenormal spectra subdirectory288. If the current plasma process deviates from Process A, then thesubroutine790 will compare the entirety of the subject current plasma process with Process B in thenormal spectra subdirectory288. Only if the current plasma process deviates from Process B will other plasma processes stored in thenormal spectra subdirectory288 be compared one at a time with the current plasma process by the plasma health/process recognition subroutine790. As in the case of theplasma health subroutine253 discussed above in relation to FIG. 21, the plasma health/process recognition subroutine790 may be configured to make all plasma processes stored in thenormal spectra subdirectory292 available for comparison with the current plasma process, or the above-noted refining criterion/criteria may be used.

The first spectra of the plasma in theprocessing chamber36 obtained for the plasma health/process recognition subroutine790 is through execution ofstep794 and which is also part of the initialization of thesubroutine790. This spectra is associated with the time t₀(hereafter “start time t₀”), and is stored along with each spectra obtained for thesubroutine790 until its analysis is completed. Any failure to retain the spectra of the current plasma process would not allow thesubroutine790 to use its “one process at a time” comparative logic.

Assume that the first plasma process stored in thenormal spectra subdirectory292 to be compared with the current plasma process being run in thechamber36 is Process A. The first step of thesubroutine790 which is repeated for each plasma process evaluated by thesubroutine790 isstep798 where a current time t_cvariable is introduced, and further where this current time t_cis set equal to the starting time t₀. Comparison of the current plasma process being run in theprocessing chamber36 with the data fromstep816 is undertaken atstep800 where the plasma health/process recognition subroutine790 is directed to proceed to thepattern recognition module370 of FIG.13. An analysis of the spectra of the plasma from theprocessing chamber36 at the current time t_cis undertaken atstep800 to determine if the pattern of this current spectra is a “match” with the relevant spectra of Process A of thenormal spectra subdirectory288. As discussed above in relation to FIG. 13, the “match determination” is effectively a comparison of the patterns of the two noted spectra to determine if the pattern of the current spectra is sufficiently similar to the pattern of the relevant spectra of Process A from thenormal spectra subdirectory288 to be considered a “match” therewith. “Relevance” in terms of which spectra of a given plasma process from thenormal spectra subdirectory288 is compared with the spectra of the plasma in thechamber36 at the current time t_cby thesubroutine790 may be determined in accordance with either the time dependency requirement or the progression dependency requirement discussed above in relation to theplasma health subroutine253 of FIG.21.

The results of the analysis fromstep800 are evaluated atstep812 of the plasma health/process recognition subroutine790. If the spectra of the plasma in theprocessing chamber36 associated with the current plasma process at the current time t_cis a “match” with the relevant spectra of Process A in thenormal spectra subdirectory288, thesubroutine790 proceeds to step802 where the results are displayed. For instance, an indication may be provided to operations personnel on the display130 (FIG.6), or by any of the other methods described above, that the plasma health/process recognition subroutine790 has determined that the current plasma process being run in theprocessing chamber36 corresponds, through the current time t_c, with Process A. It may be inaccurate and therefore inadvisable at this point in time to indicate that the plasma process currently being run in theprocessing chamber36 is definitively Process A. Specifically, the comparison of the current plasma process with thenormal spectra subdirectory288 up to this time has been limited to Process A. The spectra of the plasma in thechamber36 up through the current time t_ccould in fact also “match” the relevant spectra of one or more other plasma processes stored in thenormal spectra subdirectory288. However, this has not yet been determined as the logic employed by the plasma health/process recognition subroutine790 is to evaluate the current plasma process being run in theprocessing chamber36 against only a single plasma process in thenormal spectra subdirectory288 at a time. Therefore, at this time all that should be said is that the current plasma process is potentially Process A.

The status of the evaluation of the current plasma process is checked atstep806 of the plasma health/process recognition subroutine790. Execution ofstep806 determines if all of the data from the current plasma process has been evaluated by the subroutine790 (e.g., has all the data obtained up until the plasma goes “off” been evaluated). Any continuation of the current plasma process or a failure to have examined all of its optical emissions data will cause the plasma health/process recognition subroutine790 to proceed to step804 which causes the current time t_cto be adjusted by an increment of “n.” The magnitude of “n” defines the Analytical Time Resolution, and preferably the Preferred Analytical Time Resolution is implemented. For instance, if the start time to was set at one second (where the initial spectra reading was obtained for thesubroutine790 at step794) and the variable “n” was set at two seconds, the current time t_cupon exitingstep804 would be 3 seconds. The spectra at this new current time t_cfrom theprocessing chamber36 is then obtained for thesubroutine790 atstep808, and thesubroutine790 returns to step800 where the pattern of this new spectra is compared with the pattern of the relevant spectra of Recipe A to determine if they “match” in accordance with the foregoing.

Steps800,812,802,806,804, and808 define a loop818 which continues to be executed to compare the current plasma process being run in theprocessing chamber36 with one of the plasma processes stored in the normal spectra subdirectory288 (Process A in the subject example) until one of two conditions exists. One of these conditions is where the current plasma process has been completed and “matched” an entire plasma process stored in thenormal spectra subdirectory288. In this case, the subroutine will exit fromstep806 to step810. Control of the plasma monitoring operations may be returned bystep810 to, for instance, thestartup module202 of FIG.15.

Another condition where thesubroutine790 will exit the loop818 is when the spectra of the plasma in theprocessing chamber36 at the then current time t_cdoes not “match” with the relevant spectra of the plasma process stored in thenormal spectra subdirectory288 currently being used by the plasma health/process recognition subroutine790 (Process A in the subject example). In this case, thesubroutine790 will exit fromstep812 to step814. Step814 basically inquires as to whether each plasma process stored in thenormal spectra subdirectory288 has been compared with the current plasma process by thesubroutine790 through the loop818. If at least one plasma process stored in thenormal spectra subdirectory288 has not yet been used as a comparative standard for the current plasma process after thesubroutine790 has exited the loop818, the plasma health/process recognition subroutine790 will proceed fromstep814 to step822 where data on the next plasma process stored in thenormal spectra subdirectory288 is recalled in some manner for use by thesubroutine790. This data on a plasma process stored in thenormal spectra subdirectory288 is recalled for evaluation by thesubroutine790 against the current plasma process from the time t₀through the latest current time t_c(i.e., from the very beginning of this plasma process). That is, thesubroutine790 returns to step798 fromstep822 where the current time t_cis returned to the start time t₀, and the loop818 of thesubroutine790 is entered to evaluate the current plasma process against the next plasma process stored in thenormal spectra subdirectory288 in the above-described manner.

There will be cases where the current plasma process being run in thechamber36 does not “match” any plasma process stored in thenormal spectra subdirectory288. In this case the plasma health/process recognition subroutine790 will exitstep814 and proceed to step820. The protocol ofstep820 generally directs the plasma health/process recognition subroutine790 to determine if the current plasma recipe has encountered a known error/aberration which is stored in theabnormal spectra subdirectory292. Therefore, the plasma health/process recognition subroutine790 may include the portion of theplasma health subroutine253 of FIG. 21 which pertains to theabnormal spectra subdirectory292 and theunknown spectra subdirectory296 for that matter (i.e., starting withstep266 of thesubroutine253 and including everything thereafter).

The spectra of the current plasma process which is compared with theabnormal spectra subdirectory292 in the manner discussed in relation to theplasma health subroutine253 of FIG. 21 would be the spectra following (in time) the spectra of the last current time t_cwhich matched any of the plasma processes stored in thenormal spectra subdirectory288. Consider an example in which t₀is 1 second and “n” is 2 seconds, and further where the current plasma process “matched” with Process A up until time t₃₉, the current plasma process “matched” with Process B until time t₆₁, and the current plasma process “matched” with Process C until only time t₃. The spectra from the current plasma process being run in theprocessing chamber36 which proceeded the longest time into any of the plasma process stored in thenormal spectra subdirectory288 before any deviation was identified would be the spectra at time t₆₁from Process B. Therefore, the spectra at the time t₆₃would be that which is compared with theabnormal spectra subdirectory292 in the manner described above in relation to theplasma health subroutine253 of FIG.21.

If the plasma process identified by the plasma health/process recognition subroutine790 was a plasma recipe stored in thenormal spectra subdirectory288, a variation of thesubroutine790 may be implemented which may enhance the speed of the plasma health evaluation. Once thesubroutine790 identifies that the current plasma process being run in thechamber36 is a plasma recipe (only one of the categories of plasma processes which may be evaluated by the plasma health/process recognition subroutine790), the logic of thesubroutine790 may be modified such that thesubroutine790 would thereafter at least start its analysis of each subsequent plasma process run in thechamber36 with that plasma recipe from thenormal spectra subdirectory288 which was previously identified by thesubroutine790. If the “matching” plasma process from thenormal spectra subdirectory288 happened to be the last one which was evaluated against the current plasma process by thesubroutine790, significant plasma health evaluation time could be saved by having it be the first plasma process from thenormal spectra subdirectory288 which is compared with the next plasma process run in thechamber36. This could be particularly useful when using the plasma health/process recognition subroutine790 to evaluate plasma recipes run on acassette6 ofproduction wafers18 since the same plasma recipe is typically run on eachproduction wafer18 from a givencassette6. The ability of thesubroutine790 to distinguish between the same plasma recipe run on aqualification wafer18 versus aproduction wafer18 provides for further variations along this same line. Assume that thefirst wafer18 was actually aproduction wafer18 and that the plasma recipe was identified in the first running of this plasma recipe as being for aproduction wafer18. Eachwafer18 subsequently processed could first be checked against the plasma recipe run for aproduction wafer18 from thenormal spectra subdirectory288, and then against the plasma recipe run for aqualification wafer18 from thenormal spectra subdirectory288.

Plasma Health/Process Recognition Subroutine852—FIG.23

The plasma health/process recognition subroutine790 of FIG. 22 may be described as incorporating a “series” logic. That is, the comparison of the current spectra of the plasma in theprocessing chamber36 at the current time t_cis made in relation to only one plasma process stored in thenormal spectra subdirectory288 at a time. A plasma health/process recognition subroutine which may be used by theprocess health module252 and which proceeds with a “parallel” logic is presented in FIG.23. The plasma health/process recognition subroutine852 of FIG. 23 begins atstep854 where the Target Directory for thepattern recognition module370 of FIG. 13 is set to the normal spectra subdirectory288 (i.e., the search for “matching” spectra will initiate in the normal spectra subdirectory288). Another preliminary step of the plasma health/process recognition subroutine852 is atstep856 where a logic operator Flag₂is set to “T” for each of the plasma processes stored in thenormal spectra subdirectory288 to be evaluated through thesubroutine852. The order in which steps854 and856 are executed is not particularly important in relation to the present invention.

Data relating to the current plasma process being run on product in theprocessing chamber36 is obtained for the plasma health/process recognition subroutine852 atstep860. Included in this data is at least a spectra of the plasma within theprocessing chamber36 during the execution of a plasma process within theprocessing chamber36 at the current time t_c, which was obtained from thechamber36 over the Preferred Optical Bandwidth and at the Preferred Data Resolution. Basically, a comparison is thereafter made of the pattern of this current spectra at the current time t_cwith the relevant spectra of each plasma process stored in thenormal spectra subdirectory288 which has matched the current plasma process being run in theprocessing chamber36 up until the now current time t_c, and this comparison is made before spectra associated with a later current time t_cis compared with these plasma processes stored in the normalspectral subdirectory288. Stated another way, the current plasma process is effectively concurrently compared with each plasma process stored in thenormal spectra subdirectory288 which has “matched” the current plasma process being run in theprocessing chamber474 up until the current point in time. If the current spectra at a current time t_cdoes not match a particular plasma process stored in thenormal spectra subdirectory288, this plasma process is dropped from the list of possible plasma processes and spectra at new, later in time, current times t_care no longer compared with this plasma process. “Relevance” in terms of which spectra of a given plasma process stored in thenormal spectra subdirectory288 is compared with the spectra of the plasma in thechamber36 at the current time t_cby thesubroutine852 may be determined in accordance with either the time dependency requirement or the progression dependency requirement discussed above in relation to theplasma health subroutine253 of FIG.21. Moreover, as in the case of theplasma health subroutine253 discussed above in relation to FIG. 21, the plasma health/process recognition subroutine852 of FIG. 23 may be configured to make all plasma processes stored in thenormal spectra subdirectory292 available for comparison with the current plasma process, or thesubroutine852 may be configured such that only those certain plasma processes are made available to the plasma health/process recognition subroutine852.

The spectra at the current time t_cfromstep860 of the plasma health/process recognition subroutine852 is effectively concurrently compared with each of the relevant plasma process stored in thenormal spectra subdirectory288 the first time through the main body of the plasma health/process recognition subroutine852. The logic operator “Flag₂” associated with each such plasma process has been set to “T” at theprevious step856, so thesubroutine852 will proceed through steps864 (Process A),880 (Process B), and892 (Process “X”) to steps868 (Process A),884 (Process B), and892 (Process “X”) to where thesubroutine852 is directed to proceed to thepattern recognition module370 of FIG.13. Thepattern recognition module370 determines if the pattern of the current spectra at the current time t_cis a “match” with the relevant spectra of the subject plasma process stored in the normal spectra subdirectory288 (Process A in the case ofstep868, Process B in the case of step884, and Process “X” in the case of step896). If the current spectra at the current time t_c“matches” the relevant spectra of the subject plasma process, thesubroutine852 proceeds to step872 where the “clock” of thesubroutine852 is adjusted by increasing the current time t_cby a factor of “n”. The magnitude of “n” defines the Analytical Time Resolution. Thesubroutine852 then proceeds fromstep872 to step916 where all of the plasma processes from thenormal spectra subdirectory288 which are still a potential “match” for the current plasma process being run in theprocessing chamber36 are displayed to the appropriate personnel (e.g., ondisplay132 in FIG.6). Another spectra at this new current time t_cis then obtained for the plasma health/process recognition subroutine852 back atstep860 and the foregoing is repeated so long as the evaluation of the entirety of the current plasma process has not yet been completed by the plasma health/process recognition subroutine852 as noted in itsstep918. When the current plasma process is terminated and all spectral data has been evaluated by the plasma health/process recognition subroutine852, thesubroutine852 proceeds fromstep918 to920 where control of plasma monitoring operations may be returned to, for instance, thestartup module202 of FIG.15.

Some plasmia processes stored in thenormal spectra subdirectory288 being used by thesubroutine852 sooner or later will fail to “match” the current plasma process being run in theprocessing chamber36. That is, thepattern recognition module370 will determine that the pattern of the spectra at the current time t_cdoes not match the relevant spectra of the subject plasma process stored in thenormal spectra subdirectory288. One or more ofsteps868,884, and896 will then exit in such a manner that the logic operator “Flag₂” of their respective plasma process will be set to “F” (atstep876 for Process A, atstep888 for Process B, atstep900 for Process “X”). Any plasma process in thenormal spectra subdirectory288 having its logic: operator “Flag₂” set to “F” will no longer be compared to the current plasma process being run in theprocessing chamber36 by thepattern recognition module370 through thesubroutine852. Step868 associated with Process A will be bypassed throughstep864 when the logic operator “Flag₂” for Process A is set to “F”, step884 associated with Process B will be bypassed throughstep880 when the logic operator “Flag₂” for Process B is set to “F”, and step896 associated with Process “X” will be bypassed throughstep892 when the logic operator “Flag₂” for Recipe “X” is set to “F”.

As long as at least one plasma process stored in thenormal spectra subdirectory288 has “matched” each spectra at each new incremental time t_cfromstep860, the plasma health/process recognition subroutine852 will continue viasteps904 and912. However, if this is not the case, the plasma health/process recognition subroutine852 will exit fromstep904 to step908. The protocol of step908 is directed to determining if the current plasma process being run in theprocessing chamber36 has encountered a known error/aberration that is stored in theabnormal spectra subdirectory292. Therefore, the plasma health/process recognition subroutine852 may include the portion of theplasma health subroutine253 of FIG. 21 which pertains to theabnormal spectra subdirectory292 and theunknown spectra subdirectory296 for that matter (i.e., starting withstep266 of thesubroutine253 and including everything thereafter, but not shown). The spectra compared with theabnormal spectra subdirectory292 instep266 of theplasma health subroutine253 would be the spectra associated with the last current time t_cfromstep860 of the plasma health/process recognition subroutine852.

Now consider an example where there are three plasma processes recorded in the normal spectra subdirectory288 (e.g, Process “X” is Process C), where “n” fromstep872 is one second, wherestep860 was first executed for time t₁, where a time dependency requirement is being used by thesubroutine852 to define what is the relevant spectra from a given plasma process stored in thenormal spectra subdirectory288, and where the current plasma process being run in theprocessing chamber36 has matched each of Processes A, B, and C through the current time t₁₀(i.e., steps868 (Process A),884 (Process B), and896 (Process C) have each been executed 10 times). The “clock” of thesubroutine852 is then adjusted atstep872 to t₁₁, the spectra of the plasma in theprocessing chamber36 at this new time t₁₁is obtained for thesubroutine852 atstep860, and thesubroutine852 proceeds tosteps868,884, and896 since the logic operator “Flag₂” associated with each plasma process stored in thenormal spectra subdirectory288 is still “T” (steps864,880, and892 each exit as a “yes” condition). Assume that the current plasma process still matches Processes A and B from thenormal spectra subdirectory288 at the now current time t₁₁, but no longer matches Process C of thesubdirectory288 at the now current time t₁₁. In this case, step868 for Process A and step884 for Process B would each still exit “yes” and proceed to step872 where the clock of thesubroutine852 would be adjusted to t₁₂. The logic operator “Flag₂” for Processes A and B would also still be “T”, while step896 for Process C would exit to step900 where the logic operator “Flag₂” for Process C would be set to “F”. As such,step916 would indicate that the current plasma process was now only potentially Processes A and B.

The spectra for the new time t₁₂is obtained for thesubroutine852 atstep860 in the subject example, and thesubroutine852 proceeds to the logic operator check for each of the plasma processes stored in thenormal spectra subdirectory288. Thesubroutine852 would continue the comparison of the current plasma process being run in theprocessing chamber36 with Processes A and B in the subject example, throughsteps864 and868 for Process A and throughsteps880 and884 for Process B. However, Process C would no longer be compared with the current plasma process at the current time t₁₂sincestep892 associated with Process C would bypass its associated comparison step896, and would instead proceed to step904. Thesubroutine852 would proceed with the comparison of the current plasma process being run in theprocessing chamber36 with thenormal spectra subdirectory288 since the logic operator “Flag₂” for each of Process A and B atstep904 was still “T” at the now current time t₁₂based upon the logic fromstep904.

Now consider that the spectra at the current time t₁₂matches the spectra at this same time t₁₂for Process A (via step868), but not for Process B (via step884). The logic operator “Flag₂” for Process B would then be set to “F” atstep888. Moreover, the “clock” of thesubroutine852 would be adjusted to t₁₃atstep872, and step916 would indicate that Process A from thenormal spectra subdirectory288 was the only remaining possibility for the current plasma process being run within theprocessing chamber36. The spectra for the new current time t₁₃is obtained for thesubroutine852 atstep860 in the subject example, and thesubroutine852 proceeds to the logic operator check for each of the plasma processes stored in thenormal spectra subdirectory288 which are available to thesubroutine852. Thesubroutine852 would continue the comparison of the current plasma process being run in theprocessing chamber36 only with Process A through its associatedsteps864 and868. Plasma Processes B and C would no longer be compared with the current plasma process sincestep880 associated with Process B would bypass its comparison step884 and direct thesubroutine852 to instead proceed to step904, and sincestep892 associated with Process C would bypass its comparison step896 and direct thesubroutine852 to proceed to step904. Thesubroutine852 would proceed with the comparison of the current plasma process being run in theprocessing chamber36 with thenormal spectra subdirectory288 viastep904 since the logic operator “Flag₂” for Process A is still “T” at the now current time t₁₃.

Completion of an entire plasma process which was run in theprocessing chamber36 while matching at least one plasma process stored in the normal spectra subdirectory288 (Process A in the above-noted example) will cause thesubroutine852 to exit fromstep918 and proceed to step920. Control of the plasma monitoring operations may then be returned bystep920 of the plasma health/process recognition subroutine852 to, for instance, thestartup module202 of FIGS.13.

Plasma Health/Process Recognition Subroutine924—FIG.24

Yet another embodiment of a plasma health subroutine which may be used by theplasma health module252 is illustrated in FIG.24. The plasma health/process recognition subroutine924 is generally directed to achieving an increase in the speed of the comparison between the current plasma process being run in theprocessing chamber36 and thenormal spectra subdirectory288 by at least initially limiting the search within thesubdirectory288 to a single plasma process of thesubdirectory288. In this regard, personnel are allowed to indicate which plasma process is to be run in theprocessing chamber36. For instance, thedata entry device60 for the main control unit58 (FIG. 1) may be used to select a plasma recipe to be run from a list of plasma recipes on thedisplay130. Thestartup module202 could prompt personnel to input the recipe if desired through execution ofstep230 of thestartup routine203 of FIG.16. More typically, the recipe to be run on a certain lot ofwafers18 will be input somewhere in the fabrication facility (e.g., main control panel), and when thechamber36 “reads” the lot from thiswafer18, the plasma recipe which was input in association with this lot will be automatically run.

Once the selection has been made atstep928, the plasma health/process recognition subroutine924 proceeds to step932 to confirm that this plasma process is in fact stored in thenormal spectra subdirectory288. Information in theprocess genus field322h(e.g., plasma recipe, plasma clean, conditioning wafer) and/or process species typefield322c(e.g., a specific plasma recipe) of FIG. 12A may be used bystep932 of the plasma health/process recognition subroutine924 of FIG.24.

A spectra of the plasma in theprocessing chamber36 at the current time t_cis obtained for thesubroutine924 atstep940 if the process selected or input atstep928 was located in thenormal spectra subdirectory288 through execution ofstep936 This current spectra is compared with the relevant spectra of the selected plasma process in thenormal spectra subdirectory288. The comparison atstep944 of therecipe recognition subroutine924 determines if the pattern of the current spectra at the current time t_c(from the current plasma process being run in the processing chamber36) is a “match” with the relevant spectra of the selected plasma process stored in thenormal spectra subdirectory288. “Matches” in accordance withstep944 may be evaluated through thepattern recognition module370 of FIG.15. “Relevance” in terms of which spectra of the selected plasma process in thenormal subdirectory288 is compared with the spectra of the plasma in thechamber36 at the current time t_cby thepattern recognition module370 may be determined in accordance with either the time dependency requirement or the progression dependency requirement discussed above in relation to theplasma health subroutine253 of FIG.21.

A number of actions are undertaken if there is a “match” atstep944 of the plasma health/process recognition subroutine924, and the order of execution of these steps is relatively insignificant in relation to the present invention. Initially, the “clock” of the plasma health/process recognition subroutine924 is adjusted by a factor of “n” (which defines the Analytical Time Resolution) atstep948 to provide a new current time t_c. Moreover, the identification of the plasma process of the current operation is displayed to appropriate personnel through execution of step964 (e.g., on thedisplay130 of FIG.6). Finally, step958 inquires as to whether all of the data of the current plasma process has been evaluated.

Steps940,944,948,964, and958 of the plasma health/process recognition subroutine924 are repeated so long as the current plasma process being run on product within theprocessing chamber36 is a “match” with the selected plasma process fromstep928 and until all of the data on the current plasma process has been evaluated by thesubroutine924, in which case control of plasma monitoring operations is transferred to, for instance, thestartup module202 of FIG.15 through execution ofstep962 of thesubroutine924. Any failure of the current plasma process to “match” the process fromstep928 will cause the plasma health/process recognition subroutine924 to proceed fromstep944 to anoptions module952. It should be noted that theoptions module952 may also be accessed if the plasma process selected instep928 is not initially located in the normal spectra subdirectory299. How the plasma health/process recognition subroutine924 proceeds in these type of circumstances will likely be determined by the operator of the facility which implements thewafer production system2. Access to the entirety of thenormal spectra subdirectory288 for comparison against the current plasma process may be implemented bystep960 including protocol to call theplasma health subroutine253 of FIG. 21, the plasma health/process recognition subroutine790 of FIG. 22, or the plasma health/process recognition subroutine852 of FIG.23. Notification of the deviation of the current plasma process from the process selected instep928 of the plasma health/process recognition subroutine924 may be provided through execution ofstep956 which calls theprocess alert module428 discussed above in relation to FIG.14 and which may also offer one or more protocols in relation to this condition if the process control feature is enabled atstep436 of theprocess subroutine432. Other options such as allowing the present plasma process to be terminated (even though it may be a valid plasma process) may also be provided (not shown).

A variation of thesubroutine924 relates to the fact that the same plasma recipe is typically run on anentire cassette6, and that thecassette6 may have one ormore qualification wafers18 in with theproduction wafers18. Even though the same plasma recipe is run on thesewafers18, certain differences between theproduction wafers18 and the qualification wafer(s)18 may produce differences in their respective spectral patterns. The logic of thesubroutine924 may be to first compare the current plasma process against the same plasma recipe associated with aproduction wafer18 in thenormal spectra subdirectory288, and to compare the current plasma recipe against the same plasma recipe associated with aqualification wafer18 in thenormal spectra subdirectory288 only if the current plasma recipe does not “look” like a plasma recipe for aproduction wafer18. Moreover, entries for both production and qualification wafers of this same plasma recipe could be simultaneously evaluated in the manner presented above in relation to the plasma health/process recognition subroutine852 of FIG. 23 when the plasma recipe is input to thesubroutine924.

Plasma Health/ProcessStep Recognition Subroutine972—FIG.25

Another embodiment of a subroutine which may be used by theplasma health module252 is presented in FIG.25. Not only does thesubroutine972 of FIG. 25 monitor or evaluate the health of the plasma from a plasma process being run in theprocessing chamber36, but thesubroutine972 is also able to identify the current plasma step of the current plasma process being run in theprocessing chamber36. As such, thesubroutine972 is characterized as a plasma health/processstep recognition subroutine972. Two factors are key to providing this plasma step identification function. One is that the steps of the subject plasma process actually differ sufficiently in relation to their subject spectra such that they can be distinguished as is the case of the multi-step recipe illustrated in FIGS.17A-C above. Another is that the identify of the plasma step be associated in some manner with its corresponding spectra, such as through inputting information to theplasma step field322ediscussed above in relation to FIG.12A.

The plasma health/processstep recognition subroutine972 proceeds with a “parallel” logic and in the same manner as the plasma health/process recognition subroutine852 of FIG.23. The plasma health/processstep recognition subroutine972 of FIG. 25 begins atstep976 where the Target Directory for thepattern recognition module370 of FIG. 13 is set to the normal spectra subdirectory288 (i.e., the search for “matching” spectra will initiate in the normal spectra subdirectory288). Another preliminary step of the plasma health/processstep recognition subroutine972 is atstep980 where a logic operator Flag, is set to “T” for each of the plasma processes stored in thenormal spectra subdirectory288. The order in which steps976 and980 are executed is not particularly important to the present invention.

Data relating to the current plasma process being run in theprocessing chamber36 is obtained for thesubroutine972 at step984. Included in this data is a spectra of the plasma within theprocessing chamber36 during execution of a plasma process being run within theprocessing chamber36 at the current time t_c, Basically, a comparison is thereafter made of the pattern of this current spectra at the current time t_cwith the pattern of the relevant spectra of each plasma process stored in thenormal spectra subdirectory288 which has matched the current plasma process up until the now current time t_cand which has been made available to thesubroutine972. This comparison is made before the pattern of spectra associated with a later in time current time t_cis compared with these same processes stored in thenormal spectra subdirectory288. Stated another way, the current plasma process is effectively concurrently compared with each plasma process stored in thenormal spectra subdirectory288 which has “matched” the current plasma process up until the current point in time and which is made available to thesubroutine972. If at any time the spectra at the current time t_cfrom the current plasma process does not match a particular process in thenormal spectra subdirectory288, this plasma process is dropped from the list of possible plasma processes and spectra at new, later in time, current times t_care no longer compared with this particular plasma process. “Relevance” in terms of which spectra of the selected plasma process is compared with the spectra of the plasma in thechamber36 at the current time t_cby thesubroutine972 may be determined in accordance with either the time dependency requirement or the progression dependency requirement discussed above in relation to theplasma health subroutine253 of FIG.21. Moreover, as in the case of theplasma health subroutine253 discussed above in relation to FIG. 21, the plasma health/processstep recognition subroutine972 of FIG. 25 may be configured to make all plasma processes stored in thenormal spectra subdirectory292 available for comparison with the current plasma process, or thesubroutine972 may be configured such that only those certain of these plasma processes are made available to thesubroutine972 in any of the above-described manners.

The spectra at the current time t_cfrom step984 of the plasma health/processstep recognition subroutine972 is effectively concurrently compared with each plasma process stored in thenormal spectra subdirectory288 the first time through the main body of thesubroutine972. The logic operator “Flag₃” associated with each plasma process has been set to “T” at theprevious step980, so thesubroutine972 will proceed through steps988 (Process A),996 (Process B), and1004 (Process “X”) to steps992 (Process A),1000 (Process B), and1008 (Process “X”) to where thesubroutine972 is directed to proceed to thepattern recognition module370 of FIG.13. Thepattern recognition module370 determines if the pattern of the current spectra at the current time t_cis a “match” with the pattern of the relevant spectra of the subject plasma process stored in the normal spectra subdirectory288 (Process A in the case ofstep992, Process B in the case ofstep1000, and Process “X” in the case of step1008). If the pattern of the current spectra at the current time t_c“matches” the pattern of the relevant spectra of the subject plasma process from thesubdirectory288, thesubroutine972 proceeds to step1012 where the clock of thesubroutine972 is adjusted by increasing the current time t_cby a factor of “n”. The magnitude of “n” defines the Analytical Time Resolution (i.e., what portion of the collected data is actually analyzed). Thesubroutine972 then proceeds fromstep1012 to step1016 where all of the plasma processes from thenormal spectra subdirectory288 which are still a potential “match” for the current plasma process being run in theprocessing chamber36 are displayed to the appropriate personnel (e.g., ondisplay130 in FIG.6). Moreover, the specific process step, if any, of each of these potential plasma processes is also displayed atstep1016. Another spectra at the new current time t_cis then obtained back at step984 and the foregoing is repeated so long as the evaluation of the entirety of the current plasma process has not yet been completed by the plasma health/processstep recognition subroutine972 as noted instep1044. When the plasma process is terminated and all spectral data has been evaluated by the plasma health/processstep recognition subroutine972, thesubroutine972 will proceed fromstep1044 to1048 where control of plasma monitoring operations may be returned to, for instance, thestartup module202 of FIG.15.

Some plasma processes in thenormal spectra subdirectory288 sooner or later will fail to “match” the current plasma process being run in theprocessing chamber36. That is, thepattern recognition module370 will determine that the pattern of the spectra at the current time t_cdoes not match the pattern of the relevant spectra of the subject plasma process stored in thenormal spectra subdirectory288. One or more ofsteps992,1000, and1008 will then exit in such a manner that the logic operator “Flag₃” of their respective plasma process will be set to “F” (atstep1020 for Process A, atstep1024 for Process B, atstep1028 for Process “X”). Any plasma process stored in thenormal spectra subdirectory288 in which its logic operator “Flag₃” is set to “F” will no longer be compared to the current plasma process by thepattern recognition module370. Step992 associated with Process A will be bypassed throughstep988 when the logic operator “Flag₃” for Process A is set to “F”,step1000 associated with Process B will be bypassed throughstep996 when the logic operator “Flag₃” for Process B is set to “F”, and step1008 associated with Process “X” will be bypassed throughstep1004 when the logic operator “Flag₃” for Process “X” is set to “F”.

As long as at least one plasma process from thenormal spectra subdirectory288 has “matched” each spectra at each new incremental time t_cfrom step984, the plasma health/processstep recognition subroutine972 will continue viasteps1032 and1036. However, if this is not the case, thesubroutine972 will proceed fromstep1032 to step1040. The protocol ofstep1040 is directed to determining if the current plasma process being run in theprocessing chamber36 has encountered a known error/aberration that is stored in theabnormal spectra subdirectory292. Therefore, the plasma health/process recognition subroutine972 may include the portion of theplasma health subroutine253 of FIG. 21 which pertains to theabnormal spectra subdirectory292 and the unknown spectra subdirectory296 (i.e., starting withstep266 of thesubroutine253 and including everything thereafter). The spectra compared with theabnormal spectra subdirectory292 would then be the spectra associated with the last current time t_cfrom step984 of the plasma health/processstep recognition subroutine972.

Exemplary “Aging Chamber” Spectra—FIGS.26A-C

The primary purpose of theplasma health module252 is to monitor the health of the plasma used to execute a plasma process within thechamber36. As noted above, the running of plasma processes on product in thechamber36 will eventually start to adversely impact its performance. This “aging” chamber condition is often, if not always, reflected by the spectra of the plasma in thechamber36 during a plasma process. How the pattern of the spectra of the plasma may change over time as thechamber36 “ages” is illustrated in FIGS. 26A-C.

FIG. 26A presents aspectra1052 of an exemplary plasma when theprocessing chamber36 is in a clean condition and while running a corresponding plasma recipe on product (e.g., a “healthy” plasma). FIG. 26B presents aspectra1060 of this same exemplary plasma after a number of plasma recipes have been conducted within theprocessing chamber36 and while this same plasma recipe is actually being run on product in thechamber36. Although the plasma recipes run in thechamber36 between the times associated with FIG. 26A and 26B have started to age thechamber36, the interior of thechamber36 has not sufficiently degraded the health of plasma to the point where thechamber36 needs to be cleaned. Finally, FIG. 26C presents aspectra1068 of this same exemplary plasma, during the running of the same plasma recipe on product in thesame chamber36 as presented in FIGS. 26A-B, and where the running of the previous plasma recipes on product in theprocessing chamber36 has further deteriorated the condition of the interior of theprocessing chamber36. Thisspectra1068 may be selected by the operator of the facility implementing thewafer production system2 as being indicative of thechamber36 being in condition for cleaning (e.g., “dirty/unhealthy plasma”, a “dirty chamber” condition). For instance, if the product processed in thechamber36 was defective in some manner, and if analysis of the product traced the cause back to the condition of thechamber36, spectral data from the run associated with this defective product may be selected as being indicative of the subject condition. However, it should be appreciated that it may be desirable to have a dirty chamber condition identified before it starts ruining product processed in thechamber36. That is, it may be desirable to identify a trend which is that the condition of thechamber36 is degrading and associate this trend with a dirty chamber condition such that product is not lost because of a dirty chamber condition. This may be implemented by associating spectral data from runs previous to that where the defective product was produced with a dirty chamber condition (even though the product produced thereby was not yet defective). Each of thespectra1052,1060, and1068 are characterized by a number of peaks1056,1064, and1072, respectively, of varying intensities (plotted along the “y” axis and expressed in “counts” which is reflective of the intensity level) at various wavelengths (plotted along the “x” axis in nanometers). A comparison of thespectra1052,1060, and1068 reveals that their associated patterns are in fact different, including without limitation as follows: 1) at about the 440 nanometer wavelength region, peak1056ain thespectra1052 of FIG. 26A has an intensity of about 3,300, peak1064ain thespectra1060 of FIG. 26B has an intensity of about 3,300, and peak1072ain thespectra1068 of FIG. 26C has an intensity of about 2,700; 2) at about the 525 nanometer wavelength region, peak1056bin thespectra1052 of FIG. 26A has an intensity of about 2,800, peak1064bin thespectra1060 of FIG. 26B has an intensity of about 2,900, and peak1072bin thespectra1068 of FIG. 26C has an intensity of about 2,100; 3) at about the 560 nanometer wavelength region, peak1056din thespectra1052 of FIG. 26A has an intensity of about 400, peak1064din thespectra1060 of FIG. 26B has intensity of about700, and peak1072din thespectra1068 of FIG. 26C has an intensity of about 1,200; 4) at about the 595 nanometer wavelength region, peak1056ein thespectra1052 of FIG. 26A has an intensity of about 2,100, peak1064ein thespectra1060 of FIG. 26B has intensity of about 2,000, and peak1072ein thespectra1068 of FIG. 26C has an intensity of about 2,000; and 5) at about the 625 nanometer wavelength region, there is substantially no peak in thespectra1052 of FIG. 26A, although this region has an intensity of about 200 (noise),peak1064fin thespectra1060 of FIG. 26B has intensity of about 900, and peak1072fin thespectra1068 of FIG. 26C has an intensity of about 1,500. These are but a few examples, as it should be clear that the peaks1056 ofspectra1052 of FIG. 26A, the peaks 1064 of thespectra1060 of FIG. 26B, and the peaks1072 of thespectra1068 of FIG. 26C differ as to one or more of their respective wavelengths and intensities in a manner which would allow these types of spectra to be used as a basis to determine when to clean thechamber36. That is, the distinctiveness of the patterns between thespectra1052,1060, and1068 may be utilized to apprise the appropriate personnel of the condition of theprocessing chamber36 in relation to cleaning schedules.

At least two options exist for implementing the spectra in theplasma spectra directory284 of FIG. 9 which are at least deemed indicative of achamber36 which is in need of cleaning. Spectra of plasma in a dirty chamber condition may be included in theabnormal spectra subdirectory292 of FIG.9. In this case, theplasma health subroutine253 of FIG. 21, the plasma health/process recognition subroutine790 of FIG. 22, the plasma health/process recognition subroutine852 of FIG. 23, the plasma health/process recognition subroutine924 of FIG. 24, and the plasma health/processstep recognition subroutine972 of FIG. 25 would each then include “chamber condition monitoring” capabilities in the manner presented above. How one or more spectra which is indicative of a dirty chamber condition may be obtained is as follows. Consider the situation where a plasma recipe being run on product in theprocessing chamber36 does not “match” any plasma recipe stored in thenormal spectra subdirectory288, and further does not “match” any of the known errors/aberrations stored in theabnormal spectra subdirectory292 such as discussed above in relation theplasma health subroutine253 of FIG.21. The spectra of the plasma from this plasma recipe would then be stored in theunknown spectra subdirectory296. When spectra from this plasma recipe are subsequently examined and determined to be associated with a dirty chamber condition, at least one of the spectra from this plasma recipe which did not “match” any entry in thenormal spectra subdirectory288 or theabnormal spectra subdirectory292 may be selected as being indicative of the dirty chamber condition. This spectra may then be transferred to theabnormal spectra subdirectory292 such that this spectra will be identified as a known error condition. The protocol which may be implemented (either automatically or manually) upon encountering this same condition in subsequent executions of plasma processes in thissame chamber36 may be as set forth above in the discussion of theprocess alert subroutine432 of FIG.14. One or more alerts of the dirty chamber condition may also be issued consistent with the above discussion.

Chamber Condition Module1084—FIGS.27-29

Another way of implementing spectra indicative of a dirty chamber condition is to include this data in achamber condition subdirectory300 and to utilize achamber condition module1084 which is separate from theplasma health module252. One embodiment of a subroutine which may be used to monitor the condition of thechamber36 through a comparative analysis with thechamber condition subdirectory300 is illustrated in FIG.27. There are a number of ways in which thechamber condition subroutine406 of FIG. 27 may be implemented to monitor current plasma processes being run in thechamber36. Theloop190 of theplasma health subroutine253 of FIG. 16 may include protocol in itsstep194 to call thechamber condition subroutine406 of FIG.27. That is, thesubroutine406 may be called by theplasma health subroutine253 for each execution ofstep254 of thesubroutine253 where a current spectra at the current time t_cis obtained for thesubroutine253. This same spectra may then be made available to thechamber condition subroutine406 through execution ofstep410 of thesubroutine406. The pattern of this spectra may then compared with the pattern of spectra in thechamber condition subdirectory300 at step412 (step408 of thesubroutine406 sets the Target Directory for thepattern recognition module370 of FIG. 13 to the chamber condition subdirectory300). More specifically, step412 of thechamber condition subroutine406 directs thesubroutine406 to go to thepattern recognition module370 of FIG.13. If the pattern of the spectra of the plasma in theprocessing chamber36 at the then current time t_cdoes not “match” the pattern of any spectra in thechamber condition subdirectory300, thechamber condition subroutine406 will proceed fromstep414 to step416. Control is then returned to theloop190 of theplasma health subroutine253 of FIG. 21 where the “clock” is adjusted by an increment of “n” at itsstep278. Another spectra is then obtained for thechamber condition subroutine406 at this new current time t_cwhenstep194 of theloop190 from theplasma health subroutine253 of FIG. 21 is again encountered, which calls thechamber condition subroutine406 of FIG. 27 for a repetition of the above-described analysis. Any of thesubroutines790,852,924, and972 for providing plasma health evaluations may include this type of feature (not shown).

Thechamber condition subroutine406 of FIG. 27 will continue to execute in the above-noted manner until one of two conditions is met. The first is when plasma monitoring operations are terminated through theplasma health subroutine253 of FIG. 21, such as when all of the data on the current plasma process has been evaluated and the plasma processing operations have been terminated. The second condition which will cause the termination of thechamber condition subroutine406 in the above-described manner is when thepattern recognition module370 indicates that there is a match between the spectra of the plasma in theprocessing chamber36 at the then current time t_cand at least one spectra in thechamber condition subdirectory300. In this case thesubroutine406 will proceed fromstep414 to step418 where thechamber condition subroutine406 of FIG. 44 transfers control to theprocess alert module428 where any of the actions described above in relation to a dirty chamber condition may be initiated.

Thechamber condition subroutine406 may be run in parallel with theplasma health module252 each time theplasma health module252 is accessed (e.g., by havingstep236 of thestartup routine203 also include protocol to call the chamber condition subroutine406). In this case, an additional step of adjusting the clock would have to be included, as well as a loop defined by this step as well assteps410,412, and414 in a manner similar to the other subroutines presented herein.

There are ways of determining when the interior of theprocessing chamber36 is in condition for cleaning other than comparing a pattern of a current spectra of the plasma in thechamber36 against the pattern of a spectra which has been previously associated with a dirty chamber condition. One such way is implemented by thechamber condition subroutine1088 presented in FIG.28. The premise employed by thechamber condition subroutine1088 is that the interior of theprocessing chamber36 has degraded to the point where a cleaning operation should be employed when the time required to complete any plasma step of a multiple step plasma process takes longer than a time limit previously established for the plasma step. The time required to complete a plasma step may increase a as the condition of the interior of theprocessing chamber36 degrades. As an example, a given plasma step may take 30 seconds to achieve its desired/predetermined end result in a “clean” processingchamber36, may take 50 seconds in achamber36 which is at an intermediate time in relation to a “cleaning” cycle, and may take in excess of 70 seconds under dirty chamber conditions. In some cases theprocessing chamber36 may actually degrade to the point where the result of the plasma step may never be realized. Therefore, thechamber condition subroutine1088 assumes that when a given plasma step takes longer than its associated time limit, the associated cause is the existence of a dirty chamber condition.

Since thechamber condition subroutine1088 needs to “know” the identify of both the plasma process and the process step to execute its analysis, thechamber condition subroutine1088 may be integrated in some manner with the operation of any one or more of the plasma health/process recognition subroutine790 of FIG. 22, the plasma health/process recognition subroutine852 of FIG. 23, or the plasma health/process recognition subroutine924 of FIG. 24 (e.g., by being incorporated therein, by being called simultaneously therewith). In each of these cases, once the identity of the plasma process is determined, one will know each of the particular plasma steps which are included in this plasma process. In addition, thechamber condition subroutine1088 may also be integrated in some manner with the plasma health/processstep recognition subroutine972 of FIG. 25 which identifies the current plasma step being executed in thechamber36.

Referring to FIG. 28, a maximum time limit for each plasma step, if any, of the plasma process to be run in thechamber36 should be obtained by thechamber condition subroutine1088 at itsstep1092. Personnel may manually input the maximum time limit for the subject plasma step(s) of the plasma process with thedata entry device132 for purposes ofstep1092 of thechamber condition subroutine1088. A more preferred approach is to include these time limits in the maximum total processstep time field322ffor themain data entry350 of the plasma process as stored in the normal spectra subdirectory288 (FIG.12A). The maximum time limit may be empirically determined and input to the subject maximum total processstep time field322f.Alternatively, the limits referred to instep1092 may simply coincide with a time in which the operator of the fabrication facility employing thewafer production system2 has determined is necessary to maintain a desired production rate, which would then be input to the subject maximum total processstep time field322f.Information forstep1092 of thechamber condition subroutine1088 may then be automatically retrieved from the corresponding maximum total processstep time field322fof FIG. 12A which is associated with the current plasma process step being run in thechamber36.

The amount of time required to complete each plasma step of the current plasma process being run in theprocessing chamber36 is monitored at step1096 of thechamber condition subroutine1088. A process step clock (not shown) is started once the subject plasma step is initiated and will not stop until the termination of this plasma step. Step1096 may utilize theendpoint detection module1200 to be discussed below in relation to FIGS. 52-58 to identify the endpoint of the current plasma step. A comparison is made atstep1100 of thechamber condition subroutine1088 between the time spent on the current plasma step from step1096 and its associated maximum time limit fromstep1092. So long as this limit has not yet been exceeded, thechamber condition subroutine1088 will proceed to step1108 where a determination is made as to whether the current plasma process has been terminated. Any continuation of the plasma process will allow thechamber condition subroutine1088 to continue its analysis through execution ofsteps1096 and1100 as described. When the plasma process is terminated, however, thesubroutine1088 will proceed fromstep1108 to step1112 where control of plasma monitoring operations may be returned to, for instance, thestartup module202 of FIG.15.

Thechamber condition subroutine1088 will continue to execute in the above-noted manner unless the time spent on the current plasma process step exceeds its corresponding maximum time limit. In this case thechamber condition subroutine1088 will proceed fromstep1100 to step1104 where theprocess alert module428 of FIG. 14 is called for a dirty chamber condition, and where the above-noted types of actions may then be undertaken. If theprocess alert module428 of FIG. 14 is configured to return control to the module which called themodule428, thechamber condition subroutine1088 may include a step afterstep1104 to return control of plasma monitoring operations to, for instance, thestartup module202 of FIG.15.

Another way of determining when the interior of theprocessing chamber36 is in condition for cleaning is embodied by thechamber condition subroutine1120 presented in FIG.29 and which may be implemented in the same way as thechamber condition subroutine1088 described above in relation to FIG.28. The premise employed by thechamber condition subroutine1120 is that the interior of theprocessing chamber36 has degraded to the point where a cleaning operation should be employed when the time required to complete an entire plasma process (all of the plasma steps) takes longer than a time limit previously established for completing the plasma process. The time required to complete an entire plasma process may increase as the condition of the interior of theprocessing chamber36 degrades, such as by the formation of deposits on its interior surfaces. As an example, a given plasma process may take 180 seconds to achieve its desired/predetermined result in a “clean” processingchamber36, may take 220 seconds in achamber36 which is at an intermediate time in relation to a “cleaning” cycle, and may take in excess of 300 seconds when in a dirty chamber condition. In some cases theprocessing chamber36 may actually degrade to the point where the desired end result of the plasma process may never be realized. Therefore, thechamber condition subroutine1120 assumes that when a given plasma process takes longer than its associated time limit, the associated cause is the existence of a dirty chamber condition.

Since thechamber condition subroutine1120 needs to “know” the identify of plasma process to execute its analysis, thechamber condition subroutine1120 may be integrated in some manner with the operation of any one or more plasma health/process recognition subroutine790 of FIG. 22, the plasma health/process recognition subroutine852 of FIG. 23, or the plasma health/process recognition subroutine924 of FIG. 24 (e.g., by being incorporated therein, by being called simultaneously therewith). In addition, thechamber condition subroutine1088 may also be integrated in some manner with the plasma health/processstep recognition subroutine972 of FIG. 25 which also identifies the current plasma process being executed in thechamber36, as well as the individual process step.

Referring to FIG. 29, a maximum time limit for the plasma process to be run in thechamber36 is input to thechamber condition subroutine1120 at itsstep1124. Personnel may manually input the maximum time limit for the subject plasma process with thedata entry device132 for purposes ofstep1124 of thechamber condition subroutine1120. A more preferred approach is to include this time limit in the maximum totalprocess time field322gfor themain data entry350 of the subject plasma process as stored in thenormal spectra subdirectory288. The maximum time period may be empirically determined and input to the subject maximum totalrecipe time field322g.Alternatively, the limit referred to instep1124 may simply coincide with a time in which the operator of the fabrication facility employing thewafer production system2 has determined is necessary to maintain a desired production rate, which would then be input to the subject maximum totalrecipe time field322g.Information forstep1124 may then be automatically retrieved from the corresponding maximum totalprocess time field322gof FIG. 12A which is associated with the current plasma process being run in thechamber36.

The amount of time required to complete the current plasma process being run in theprocessing chamber36 is monitored at step1128 of thechamber condition subroutine1120. A recipe clock (not shown) is started once the plasma process is initiated (e.g., when the plasma comes “on” in the chamber36) and will not stop until the plasma process is terminated (e.g., when the plasma goes “off” in the chamber36). Step1128 may also utilize theendpoint detection module1200 to be discussed below in relation to FIGS. 52-58. A comparison is made atstep1132 of thechamber condition subroutine1120 between the time spent on the current plasma process from step1128 and its associated maximum time limit fromstep1124. So long as this limit has not yet been exceeded, thechamber condition subroutine1120 will proceed to step1140 where a determination is made as to whether the current plasma process has been terminated. Any continuation of the plasma process will allow thechamber condition subroutine1120 to continue its analysis through execution ofsteps1128 and1132 in the described manner. When the plasma process is terminated, however, thesubroutine1120 will proceed fromstep1140 to step1144 where control of plasma monitoring operations may be returned to, for instance, thestartup module202 of FIG.15.

Thechamber condition subroutine1120 will continue to execute in the above-noted manner unless the time spent on the current plasma process exceeds its corresponding maximum time limit. In this case thechamber condition subroutine1120 will proceed fromstep1132 to step1136 where theprocess alert module428 of FIG. 14 is called based upon a dirty chamber condition, and where the above-noted actions may be undertaken. If theprocess alert module428 of FIG. 14 is configured to return control to the module which called themodule428, thechamber condition subroutine1120 may include a step afterstep1136 to return control of plasma monitoring operations to, for instance, thestartup module202 of FIG.15.

Exemplary Spectra of Cleaning Procedure—FIGS.30A-D

Some time after a “dirty chamber condition” has been identified in any of the above-noted manners, thechamber36 will undergo a cleaning operation of some type. FIGS. 30A-D illustrate the differences in the spectral patterns of the plasma in thechamber36 at various stages to illustrate how the currentplasma process module250 not only can evaluate the health of the subject plasma process, but how it can distinguish between different types of plasma processes through spectral analysis. FIG. 30A is an exemplary spectra of a plasma recipe being conducted in aprocessing chamber36 which has been determined to be in need of a cleaning operation. Compare thisspectra1440 of FIG. 30A with thespectra1450 of FIG. 30D which is this same plasma recipe in thissame processing chamber36, but when in a clean condition and after a conditioning wafer operation has been completed to prepare thechamber36 forprocessing production wafers18. Note the different intensities of the peaks in these twospectras1440 and1450 at the various wavelengths. There are two strong peaks around the 550 nanometer wavelength region in thespectra1440 of FIG. 30A, compared to only one strong peak in this wavelength region of thespectra1450 of FIG.30D. Also note the existence of a strong peak in thespectra1440 of FIG. 30A at about the 625 nanometer wavelength region, compared with no strong peak in thespectra1450 of FIG.30D. These differences can be readily identified to in turn identify when theprocessing chamber36 should be cleaned in at least some manner.

The cleaning protocol illustrated in FIG. 30 starts with a wet clean where thechamber36 is vented and opened such that interior surfaces of thechamber36 may be wiped with one or more appropriate solvents. A plasma cleaning operation may have been previously run in thechamber36, but may have been unable to adequately address the condition of the interior of thechamber36 to the desired degree. After the wet clean, thechamber36 is resealed and a plasma is introduced into thechamber36 with no product therein. Thespectra1444 of FIG. 30B is of an exemplary plasma within theprocessing chamber36 with no product therein at the start of such a plasma cleaning operation. Compare thisspectra1444 from FIG. 30B with thespectra1448 of FIG. 30C which is this same plasma in thissame processing chamber36, but after all of the residuals from the wet clean have been removed by the plasma clean. Note the different intensities of the peaks in these twospectras1444 and1448 at the various wavelengths, particularly between the 650 and 750 nanometer wavelength region where the spectra1444FIG. 30B has two substantial peaks and where the intensity of these same peaks are significantly reduced in thespectra1448 of FIG.30C.

Finally, compare thespectra1448 at the end of plasma clean and thespectra1450 at the end of a conditioning wafer operation. Thespectra1448 of FIG. 30C is of an exemplary plasma within theprocessing chamber36 at the end of a plasma clean of thechamber36 which has appropriately addressed the residuals of the wet clean. Compare thisspectra1448 from FIG. 30C with thespectra1450 of FIG. 30D which is this same plasma in thissame processing chamber36, but after a number of conditioning wafers have been processed in thechamber36. Note the different intensities of the peaks in these twospectras1448 and1450 at the various wavelengths, particularly between the 650 and 750 nanometer wavelength region where thespectra1448 of FIG. 30C has two peaks and where thespectra1450 of FIG. 30D has no peaks in this region.

Plasma Monitoring Assembly500—FIGS.3-36

Degradation of the interior of theprocessing chamber36 from plasma processes run therein also degrades theinner surface40 of thewindow38 on thechamber36 which is exposed to the plasma (theouter surface42 being isolated from the plasma, and thereby not affected by the plasma). Recall that data for the currentplasma process module250 in the nature of optical emissions of the plasma in thechamber38 is obtained through thewindow38. Therefore, as thewindow38 degrades, so to may the reliability of the results of the currentplasma process module250. An embodiment which addresses this condition, as well as other conditions which may adversely impact the reliability of the results provided by the currentplasma process module250, is presented in FIG. 31 in the nature of theplasma monitoring assembly500.

Theplasma monitoring assembly500 of FIG. 31 includes all of the features discussed above in relation to theplasma monitoring assembly174 of FIG.6. In this regard, theplasma monitoring assembly500 includes a plasma monitoring module560 (FIGS.31-32), which includes all of the same modules as identified above in relation to theplasma monitoring module200 of FIG.7 and which is part of thePMCU128′ (e.g., the same currentplasma process module250 and all of its sub-modules). Moreover, thespectrometer assembly506 andCCD array548 are also the same in relation to the similarly identified components of theplasma monitoring assembly174 of FIG.6. Increases in the accuracy of the evaluation of plasma processes by the currentplasma process module250 are available by obtaining the data for themodule250 through theplasma monitoring assembly500 because of its calibration capabilities. Specifically, theplasma monitoring assembly500, and more specifically itsplasma monitoring module560, includes al window monitoring orcalibration assembly552 which includes a window monitoring orcalibration module562 which is also part of theplasma monitoring module560 as illustrated in FIG.32. ThePMACU128′ is therefore the same as presented above in relation to theplasma monitoring assembly174 of FIG. 6, except that it includes this additional feature in relation to thecalibration module562. As such, the “prime” designation is utilized in relation to thePMCU128 of FIG.31.

Thecalibration assembly552 provides certain advantages for theplasma monitoring assembly500 of FIG. 31 over theplasma monitoring assembly174 of FIG.6. Thecalibration assembly552 of FIG. 31 provides two main functions. One of these functions is to calibrate the current plasma process module250 (FIGS. 7 and 32) for any wavelength shift of the spectra of the plasma within theprocessing chamber36. Wavelength shifts may be attributed to thespectrometer assembly506, but may also exist for other reasons. Another function provided by thecalibration assembly552 of FIG. 31 is to calibrate the currentplasma process module250 for any intensity shift of the spectra of the plasma within theprocessing chamber36. Intensity shifts may be due to an “aging” of thewindow38 on thechamber36 through which optical emissions are obtained, but may also be due to other conditions. Since thewindow478 in FIG. 31 has a different configuration than thewindow38 presented in FIG. 6, a different reference numeral is used not only for the window, but for the processing chamber as well. Therefore, the processing chamber of FIG. 31 identified by thereference number474 may be used in place of one or more of thechambers36 presented in FIG.1.

Thecalibration assembly552 of FIG. 31 includes acalibration light source556 which is operatively interfaced with thewindow478 of theprocessing chamber474 by a fiberoptic cable assembly504. In one embodiment thiscalibration light556 source actually has a firstcalibration light source556aand a secondcalibration light source556b.The firstcalibration light source556auses a first type of light to identify wavelength shifts associated with theplasma monitoring assembly500. Intensity shifts are identified through the secondcalibration light source556bwhich uses a second type of light which is different than the first type of light. Benefits associated with the use of two different types of light for identifying wavelength and intensity shifts, respectively, will be addressed in relation to thecalibration module562 and FIGS. 40-48. Although not preferred, the same type of light could be used to identify both intensity and wavelength shifts (e.g., light having a plurality of discrete intensity peaks).

The fiberoptic cable assembly504 also operatively interfaces thewindow478 with thespectrometer assembly506 such that thewindow monitoring assembly552 is able to directly monitor the condition of thewindow478. Generally, a calibration light is directed to thewindow478 from thecalibration light source556 through the fiberoptic cable assembly504, preferably when there is no plasma within theprocessing chamber474. As such, operation of thecalibration assembly552 is not dependent in any manner on the plasma within the chamber and its data is in fact independent of any data relating to the plasma. A portion of the calibration light is reflected by thewindow478 and is directed by the fiberoptic cable assembly504 to thespectrometer assembly506.

Thespectrometer assembly506 of FIG. 31 preferably is of the solid-state type, and in one embodiment includes three individual solid state spectrometers516a-c,each of which analyzes a different wavelength region and as illustrated in FIG.33. In this case, thespectrometer516amay analyze the 246 nanometer to 570 nanometer region, thespectrometer516bmay analyze the 535 nanometer to 815 nanometer region, and thespectrometer516cmay analyze the 785 nanometer to 1014 nanometer region. The overlap between these individual spectrometers516 once again facilitates alignment of the three spectral segments and reduces the potential for losing optical emissions data at the transition zones. Each of these spectrometers516 are interconnected in effectively parallel relation by the fiberoptic cable assembly504 which is illustrated in more detail in FIG.34.

The fiberoptic cable assembly504 of FIG. 34 includes three inner cables508 surrounded by six outer cables512. Light from thecalibration light source556 is directed to thewindow478 through the outer cables512 during calibration operations to be discussed below in relation to thecalibration module562 of FIGS. 40-48, while light reflected by the window478 (as well as light from within thechamber474 during the running of plasma processes within thechamber474 for that matter, at which time thecalibration assembly552 is not activated or running) is directed to thespectrometer assembly506 through the inner cables508 of the fiberoptic cable assembly504. One inner cable508 of the fiberoptic cable assembly504 is directed to each of the spectrometers516 of thespectrometer assembly506 as illustrated in FIG.33. This data received by thespectrometer assembly506 from thecalibration light source556 is directly from theinner surface482 of thewindow474, and therefore it is proper to characterize thecalibration assembly552 as directly monitoring the condition of the inner surface of thewindow478.

One function of thecalibration assembly552 is to determine the effect, if any, that theinner surface482 of thewindow478 is having on the light being emitted from theprocessing chamber474 during plasma processes since this light is used by the currentplasma process module250. Thecalibration assembly552 also addresses the presence of theouter surface486 of thewindow478. As can be seen in FIG.31 and in FIG. 35, theouter surface486 andinner surface482 of thewindow478 are disposed in a non-parallel relationship. Stated another way and as illustrated in both FIGS. 31 and 35, theinner surface482 of thewindow478 is disposed at least substantially perpendicular to areference axis490 which coincides with the primary axis of the light from thecalibration light source556 as it is directed toward thewindow478, whereas theouter surface486 of thewindow478 is disposed in non-perpendicular relation to thisreference axis490. In one embodiment, the angle between thereference axis490 and theouter surface486 of thewindow478 is within the range of about 2° to about 45°, and in another embodiment this angle is less than the critical angle. This relative positioning of theinner surface482 andouter surface486 of thewindow478 has the effect of having that portion of the calibration light, which is reflected by theouter surface486 of thewindow478, be directed away from theaxis490 and thereby away from the inner cables508 of the fiberoptic cable assembly504 which lead to thespectrometer assembly506. Therefore, when a calibration light is sent from thecalibration light source556 to thewindow478 through the outer cables512 of the fiberoptic cable assembly504, the only significant portion of light which the inner cables508 of the fiberoptic cable assembly504 “sees” is the light which is reflected by theinner surface482 of thewindow478—not from theouter surface486 of thewindow478. Theinner surface482 of thewindow478 is that which is affected by conducting plasma processes within theprocessing chamber474, and which thereby affects the light which is emitted from theprocessing chamber474 through thewindow478. Therefore, the light received by thespectrometer assembly506 during calibration operations by thewindow monitoring assembly552 presents a more accurate depiction of the condition of theinner surface482 of thewindow486. Further enhancement may be realized by incorporating a broad band anti-reflection coating (e.g., of multiple-layer or laminated construction) on theouter surface486 of thewindow478 at least in that region where the calibration light impacts theouter surface486. These types of coatings increase the amount of the calibration light which passes through theouter surface486 of thewindow478 to theinner surface482 by reducing the amount of the calibration light which is reflected by theouter surface486.

Maintaining a proper relative positioning between the fiberoptic cable assembly504 and thewindow478 on thechamber474 is important to the operation of not only thecalibration module562, but the currentplasma process module250 as well. One way of interconnecting the fiberoptic cable assembly504 and thewindow478 is illustrated in FIG. 36 in the nature of afixture assembly1534. Thefixture assembly1534 includes awindow fixture1538 which securely retains thewindow478 and allows for detachable interconnection of the same with the processing chamber474 (e.g, via one or more threaded fasteners). A cavity orrecess1542 exists within an interior portion of thewindow fixture1538 and interfaces with theouter surface486 of thewindow478. Surfaces of thewindow fixture1538 which define therecess1542 are black anodized so as to absorb any portion of the calibration light from thecalibration light source506 which is reflected by theouter surface486 of thewindow478 during calibration operations. Light absorbing coatings could also be utilized to provide this function.

Thefixture assembly1534 also includes afiber fixture1546 which is appropriately interconnected (e.g., detachably) with the window fixture1538 (e.g., via one or more threaded fasteners). Therecess1542 in thewindow fixture1538 is thereby a closed space in the assembled condition via theouter surface486 of thewindow478 and a portion of the back side of thefiber fixture1546. Appropriate treatment of the portion of thefiber fixture1546 which closes therecess1542 may also be implemented to reduce the potential for that portion of the calibration light which is reflected by theouter surface486 of thewindow478 interfering with that portion of the calibration light which is reflected by theinner surface482 of thewindow478 and provided to thespectrometer assembly506 via the fiberoptic cable assembly504.

The fiberoptic cable assembly504 is removably or detachably interconnected with thefiber fixture1546 by afiber fixture coupling1554 on thefiber fixture1546 and acable coupling1558 on an end of the fiberoptic cable assembly504 which houses both the inner cables508 and the outer cables512 (FIG.34). The ends of the inner cables508 and outer cables512 project toward theouter surface486 of thewindow478 in axial alignment with aport1550 which extends through thefiber fixture1546 to intersect therecess1542 in thewindow fixture1538. Therefore, calibration light from thecalibration light source556 is directed through the outer cables512, through theport1550 in thefiber fixture1546, through therecess1542 in thewindow fixture1538, and to theouter surface486 of thewindow478. Calibration light which is reflected by theinner surface482 of thewindow478 travels through therecess1542 in thewindow fixture1538, through theport1550 in thefiber fixture1546, into the inner cables508 of the fiberoptic cable assembly504, and to thespectrometer assembly506. The particular manner in which calibration operations are undertaken using the foregoing arrangement is addressed in relation to thecalibration module562 of FIGS. 40-48.

Plasma Monitoring Assembly700—FIGS.37-39

Another embodiment of a plasma monitoring assembly which also reduces the potential for that portion of the calibration light which is reflected by the outer surface of the window interfering with that portion of the calibration light which is reflected by the inner surface of the window and provided to the calibration module562 (FIG. 32) is illustrated in FIG.37. This embodiment may be used in place of theplasma monitoring assembly500 of FIG. 31 although the configuration presented in FIG. 31 is more preferred. Theplasma monitoring assembly700 of FIG. 37 generally includes acalibration assembly726, the calibration module562 (FIG. 32) which is part of thePMCU128′ (FIG.1 and32), and the plasma monitoring module560 (FIGS.32 and37). Thespectrometer assembly712,CCD array716, andcalibration light source728 are the same in relation to the similarly identified components of the FIG. 31 embodiment. Consequently, theplasma monitoring assembly700 of FIG. 37 differs from theplasma monitoring assembly500 of FIG. 31 principally in relation to the optical arrangement of the calibrating componentry.

One function of thecalibration assembly726 is to determine the effect, if any, that theinner surface40 of thewindow38 is having on the light being emitted from theprocessing chamber36. Thecalibration assembly726 also addresses the presence of theouter surface42 of thewindow38. As illustrated in FIGS. 37-38, thewindow38 on theprocessing chamber36 includes aninner surface40 andouter surface42 which are disposed in substantially parallel relation (as illustrated in FIGS.1 and6). In order to reduce the effects that light reflected by theouter surface42 will have on thecalibration module562, thecalibration assembly726 uses afiber optic cable704 to operatively interface thecalibration light source728 and thewindow38, and anotherfiber optic cable708 to operatively interface thewindow38 and thespectrometer assembly712. Thefiber optic cable704 is disposed to have light from thecalibration light source728 impact theouter surface42 of thewindow38 at an angle other than perpendicular, and thefiber optic cable708 is disposed to receive only that portion of the calibration light which is reflected by theinner surface40 of thewindow38 and not any light which is reflected by theouter surface42 of thewindow38. This is illustrated in FIG. 38 where theaxis732 corresponds with the orientation of the light from thecalibration light source728 as it impacts thewindow38, where theaxis736 corresponds with that portion of the calibration light which will be reflected by theouter surface42 of thewindow38, and where theaxis740 corresponds with that portion of the calibration light which will be reflected by theinner surface40 of the window38 (refraction not being shown). In one embodiment, the angle α between theaxis732 and theouter surface42 of thewindow38 ranges from about 10° to about 70°, the angle θ between theaxis740 and theinner surface40 of thewindow38 ranges from about 10° to about 70°, and theaxes736 and740 are offset by an amount ranging from about 2 mm to about 20 mm.

The above-noted relative positioning of theinner surface40 of thewindow38, theouter surface42 of thewindow38, thefiber optic cable704, and thefiber optic cable708 has the effect of having that portion of the calibration light, which is reflected by theouter surface42 of thewindow38, be reflected in a manner so as to not enter thefiber optic cable708. Therefore, when a calibration light is sent from thecalibration light source728 to thewindow38 through thefiber optic cable704, the only significant portion of light which thefiber optic cable708 “sees” is the light which is reflected by theinner surface40 of thewindow38—not light from theouter surface42 of thewindow38. Theinner surface40 of thewindow38 is that which is affected by conducting plasma processes within theprocessing chamber36, and which thereby affects the light which is emitted from theprocessing chamber36 through thewindow38. Therefore, the light received by thespectrometer assembly712 during calibration operations by thewindow monitoring assembly700 presents a more accurate depiction of the condition of theinner surface40 of thewindow36.

Further enhancement of the arrangement presented in FIG. 37 may be realized by incorporating a broad band anti-reflection coating (e.g., of multiple-layer or laminated construction) on theouter surface42 of thewindow38 at least in that region where the calibration light impacts theouter surface42. These types of coatings increase the amount of the calibration light which passes through theouter surface42 of thewindow38 to theinner surface40 by reducing the amount of the calibration light which is reflected by theouter surface42. In fact, the embodiment presented in FIG. 31 could be used with the type of window presented in the FIG. 37 embodiment where the above-noted coating is included on thewindow38 and with the ends of the fiberoptic cable assembly504 projecting toward thewindow38 to form at least a substantially perpendicular angle relative to both theouter surface42 and theinner surface40. This arrangement is less preferable in that notwithstanding the presence of an anti-reflection coating on theouter surface42 of thewindow38, some portion of the calibration light will still be reflected back toward the inner cables508 of the fiberoptic cable assembly504 in this variation (not shown). There will thereby be some interference with that portion of the calibration light which is reflected by theinner surface40 of thewindow38 back to the inner cables508.

Maintaining a proper relative positioning between thefiber optic cable704, thefiber optic cable708, and thewindow38 on thechamber36 with the FIG. 37 embodiment is important to the operation of not only thecalibration module562, but the currentplasma process module250 as well. One way of interconnecting the foregoing is illustrated in FIG. 39 in the nature of afixture assembly1564. Thefixture assembly1564 includes awindow fixture1568 which securely retains thewindow38 and allows for interconnection of the same with the processing chamber36 (e.g, via one or more threaded fasteners). A cavity orrecess1572 exists within an interior portion of thewindow fixture1568 and interfaces with theouter surface42 of thewindow38.

Thefixture assembly1564 also includes afiber fixture1576 which is appropriately interconnected with the window fixture1568 (e.g., via one or more threaded fasteners). Therecess1572 in thewindow fixture1568 is thereby a closed space in the assembled condition via theouter surface42 of thewindow38 and a portion of the back side of thefiber fixture1576. Each of thefiber optic cables704 and708 are removably interconnected with thefiber fixture1576. Afiber fixture coupling1580bon thefiber fixture1576 is disposed in the proper orientation for establishing an appropriate interconnection with thefiber optic cable704 from thecalibration light source704, while afiber fixture coupling1580aon thefiber fixture1576 is disposed in the proper orientation for establishing an appropriate interconnection with thefiber optic cable708 leading to thespectrometer assembly712. The ends of thefiber optic cables704 and708 project toward theouter surface42 of thewindow38 at the proper angle and in axial alignment with aport1584aand1584b,respectively. Theports1584aand1584beach extend through thefiber fixture1576 to intersect therecess1572 in thewindow fixture1568. Therefore, calibration light from thecalibration light source728 is directed through thefiber optic cable704, through theport1584bin thefiber fixture1576, through therecess1572 in thewindow fixture1568, and to theouter surface42 of thewindow478. Calibration light which is reflected by theinner surface40 of thewindow38 travels through therecess1572 in thewindow fixture1568, through theport1584ain thefiber fixture1576, into thefiber optic cable708, and to thespectrometer assembly712. The particular manner in which calibration operations are undertaken using the foregoing arrangement is addressed in relation to thecalibration module562 of FIGS. 40-48.

Calibration Module562—FIGS.40-48

Both thecalibration assembly552 of FIG.31 and thecalibration assembly726 of FIG. 37 include thecalibration module562 which is illustrated in FIG.40. Thismodule562 may be used to calibrate the output from each of thespectrometer assembly506 of FIG.31 and thespectrometer assembly712 of FIG.37. For convenience, the discussion will continue with regard to only thespectrometer assembly506 of FIG. 31, although such will be equally applicable to thespectrometer assembly712 of FIG. 37 as well.

Output from thespectrometer assembly506 may have a tendency to “drift” over the life of thespectrometer assembly506 due to various factors such as temperature. Any drifting of the output from thespectrometer assembly506 will cause a wavelength shift in the spectra which is evaluated by the currentplasma process module250. An example of drifting would be that a peak in a spectra from theprocessing chamber474 which is actually at the490 nanometer wavelength, may appear at the491 nanometer wavelength from the output of thespectrometer assembly506 due to this drifting. Moreover, thewindow478 may have an effect on the spectra of the plasma from the current plasma process which is passing through thewindow474 to thespectrometer assembly506, namely by providing an intensity shift in one or more regions of the spectra representative of the plasma in theprocessing chamber474. Failure to address either of these conditions may adversely affect the performance of the currentplasma process module250.

One embodiment of thecalibration module562 of FIG. 32 is presented in more detail in FIG.40 and accounts for both the wavelength and intensity shift conditions noted above. Thecalibration module562 includes acalibration routine564 which is run without any plasma in theprocessing chamber474 and which is run at times determined by the facility using the wafer production system2 (e.g., once a day, on every shift change). Instructions may be included in thecalibration routine564 to detect the existence of plasma in theprocessing chamber474 in any of the above-noted manners, and to exit thecalibration routine564 if any such plasma is detected (not shown). Step568 of thecalibration routine564 directs thecalibration light source556 to send the calibration light to thewindow478. Thereafter,step572 directs thecalibration routine564 to proceed to an appropriate calibration subroutine.

The calibration subroutine referred to instep572 of FIG. 40 may include thecalibration subroutine576 which is illustrated in FIG.41. Generally, thecalibration subroutine576 is directed toward making at least one adjustment in relation to thespectrometer assembly506 to compensate for a wavelength shift associated with the spectral data obtained through thewindow478. A comparison is undertaken bystep580 of thecalibration subroutine576 between the spectra of the calibration light from thecalibration light source556 which is directed to thewindow478 and the spectra of the calibration light which is reflected from theinner surface482 of thewindow478 and provided to thespectrometer assembly506. Since no plasma exists in theprocessing chamber474 during calibration operations, the light received by thespectrometer assembly506 should be limited to that portion of the calibration light which is reflected by theinner surface482 of thewindow478 which is therefore a direct monitoring of a condition of thewindow478, specifically itsinner surface482. Any wavelength shift from the comparison ofstep580 will be noted instep584 of thecalibration subroutine576, an adjustment will be made in relation to theplasma monitoring assembly500 to account for this shift at step588 of thesubroutine576, and control of the plasma monitoring operations will be turned over to thestartup module202 of FIG. 32 bystep592 of thecalibration subroutine576.

The comparison of the subject spectra atstep580 of thecalibration subroutine576 and the identification of any wavelength shift instep584 of thesubroutine576 may be implemented in the following manner. The spectra of the calibration light which is sent to thewindow478 may be obtained from thecalibration light subdirectory310 of FIG.9. This spectra is analyzed to identify the location of one or more of the intensity peaks in this spectra and possibly the relative wavelength positioning of these intensity peaks. A “peak” may be equated as any portion of the spectra with an intensity greater than a predetermined amount (e.g., at least about 100 intensity units) which exists over a predetermined wavelength range (e.g., no more than about 2 nanometers). Therefore, the above-referenced analysis of the spectra of calibration light sent to thewindow478 may simply entail noting the intensity over at least a portion of this spectra using an appropriate wavelength resolution. For example, one or more peaks in this spectra may be identified by noting the intensity of the spectra of the calibration light from thecalibration light subdirectory310 at every 0.5 nanometer increment over at least a portion of this spectra to identify any intensity peaks as defined above. After identifying these peaks, it may be beneficial to then note their relative wavelength positionings.

Any intensity peaks in the calibration light which is sent to thewindow478 which are identified in the above-noted manner should appear at the same wavelength and “same” intensity level (taking into consideration certain principles of optics noted below and assuming that there is no intensity shift due to thewindow478 also noted below) in that portion of the calibration light which is reflected by theinner surface482 of thewindow478 if there is no wavelength shift. Since the location and intensity of one or more peaks was noted for the calibration light that was sent to thewindow478, the amount of wavelength shift may be identified simply by finding these same peaks in the portion of the calibration light that is reflected by theinner surface482 of thewindow478 and noting any corresponding wavelength shift. Although the intensity of the peaks alone may be sufficient to identify the corresponding peaks in that portion of the calibration light that is reflected by theinner surface482 of the window478 (e.g., by looking for the largest peak around a certain wavelength), the relative positioning of the identified peaks may be desirable/required in some cases.

The wavelength shift concept addressed in relation to the calibration subroutine574 of FIG. 41 is further addressed in relation to FIGS. 42-43. One embodiment of aspectra672 of a calibration light which is appropriate for identifying a wavelength shift is illustrated in FIG. 42, and this calibration light may be used by thecalibration light source556 of the calibration assembly552 (FIG. 31) and by thecalibration subroutine576 of FIG.41. Thespectra672 is characterized by a plurality of discrete intensity peaks674 of varying intensity which is desirable for detecting a wavelength shift, and in one embodiment is from a mercury light source and comprises the above-noted firstcalibration light source556a.Other light sources having these characteristics may be utilized as well. “Intensity” is plotted along the “y” axis and is expressed in “counts” which is reflective of the intensity level, while “wavelength” is plotted along the “x” axis and is illustrated in nanometers. FIG. 42 depicts the actual pattern of calibration light which is sent to thewindow478 in the subject example. One embodiment of aspectra676 which is output by thespectrometer assembly506, due to a reflection of the calibration light represented in FIG. 42 from theinner surface482 of a “clean”window478 on theprocessing chamber474, is presented in FIG. 43 (when thewindow478 has not yet been exposed C to any plasma processes). Thespectra676 is characterized by a plurality of peaks678 of varying intensity, with “intensity” again being plotted along the “y” axis and expressed in “counts” which is reflective of the intensity level, and with “wavelength” again being plotted along the “x” axis in nanometers.

Initially, the comparison of thespectra672 of FIG. 42 with thespectra676 of FIG. 43 indicates that their respective intensities vary rather significantly. This is due to certain principles of optics. Generally, materials which would typically be used to form thewindow478 will reflect about 6% of the calibration light which is directed toward thewindow478. More specifically, 6% of the calibration light which is originally sent to thewindow478 from thecalibration light source556 will be reflected by theouter surface486 of thewindow478 along theaxis494 illustrated in FIG. 35, and the remaining 94% of the light will continue through thewindow478. When this light encounters theinner surface482 of thewindow478, 6% of this light will be reflected by theinner surface482 back toward theouter surface486, and the remaining 94% of this light will then enter the processing chamber474 (i.e., 5.64%). This light reflected by theinner surface482 will once again hit theouter surface486 which will reflect 6% of this light back toward thechamber474, while 94% will pass through theouter surface486. Therefore, only about 5.3% of the light originally sent by the calibration light source445 for calibration of the output of thespectrometer assembly506 will actually enter the inner cables508 of the fiberoptic cable assembly504 and be provided to thespectrometer assembly506. This principle of optics thereby accounts for the difference in intensity between thespectra676 of FIG.43 and thespectra672 of FIG. 42, and would have to be taken into account in the calibration procedure.

Thespectra676 of FIG. 43 may be compared with thespectra672 of FIG. 42 in the above-described manner to determine if the output from thespectrometer assembly506 needs to be calibrated for a wavelength shift. The peaks678 of thespectra676 of FIG. 43 should appear at the same wavelengths as the corresponding peaks674 of thespectra672 of FIG.43. For example, the peak678afrom FIG.43 and the peak674afrom FIG. 42 should be at the same wavelength, thepeak678bfrom FIG.43 and thepeak674bfrom FIG. 42 should be at the same wavelength, thepeak678cfrom FIG.43 and thepeak674cfrom FIG. 42 should be at the same wavelength, and so forth. In the illustrated example, the corresponding peaks from thespectras676 and672 are properly aligned or are at the same wavelengths. Therefore, no wavelength shift would be identified bysteps580 and584 of thecalibration subroutine576 of FIG.41. Moreover, with no wavelength shift being identified through execution ofstep584 of thecalibration subroutine576, no adjustment in relation to theplasma monitoring assembly500 is undertaken through execution of step588 of thesubroutine576. Control of the plasma monitoring operations is then transferred to, for instance, thestartup module202 of FIG. 15 by execution ofstep592 of thesubroutine576.

Having the peaks of thespectra672 and676 be at the exact same wavelength improves the accuracy of the currentplasma process module250. If the wavelength shift limit associated withsteps580 and584 is “0” and if the peaks of the noted spectra are offset by even a small amount, the adjustment made in relation to theplasma monitoring assembly500 at step588 will be correspondingly “small.” The amount of wavelength shift which will initiate an adjustment in relation to theplasma monitoring assembly500, however, need not be zero. That is, a certain amount of wavelength shift may be tolerated before an adjustment is made in relation to theplasma monitoring assembly500 at step588. For instance, in one embodiment no adjustments are made in relation to theplasma monitoring assembly500 unless a wavelength shift of at least a certain amount is identified bysteps580 and584 of the subroutine576 (e.g., a wavelength shift of at least about 0.25 nanometers is required before any adjustment is made).

If more than the allowed amount wavelength shift is identified instep584 of thecalibration subroutine576 of FIG. 41, as noted step588 of thesubroutine576 provides for the adjustment in relation to theplasma monitoring assembly500. At least two options exist for this “making at least adjustment in relation to theplasma monitoring assembly500.” One alternative is to physically adjust thespectrometer assembly506 if such is of the scanning type. FIG. 44 presents one embodiment of thespectrometer516a′ which is of the scanning-type. Thespectrometer516a′ includes anaperture520 through which light from theinner cable508aof thefiber optic cable504 assembly enters thespectrometer516a′. Light passing through theaperture520 is reflected by amirror524 onto agrating532. Both themirror524 and grating532 may be mounted for pivotal movement through amirror pivot528 and a grating pivot536, respectively. Movement of themirror524 and the grating532 would be affected by themotor540 and thegearbox544. Themotor540 is operatively interfaced with thePMCU128′. Movement of one or more of themirror524 and grating532 may be utilized such that the peaks678 of thespectra676 in FIG. 43 will be appropriately aligned with the corresponding peaks674 of thespectra672 in FIG.42. Thereafter, the presence of the wavelength shift being caused by thespectrometer assembly506 will not adversely affect the reliability of the results provided by the currentplasma process module250 since the wavelength shift has now been alleviated.

Another way to make the adjustment referred to in step588 of thecalibration subroutine576 of FIG. 41 is to “regression fit” the data in some way. This option may be used regardless of what type of spectrometer is implemented for thespectrometer assembly506. Consider the example where the comparison between thespectra676 of FIG.43 and thespectra672 of FIG. 42 identifies the existence of a wavelength shift (not shown since the corresponding peaks674 and678 are properly aligned). The output from thespectrometer assembly506 may be regression fit to account for the noted wavelength shift. It should be appreciated that if a solidstate spectrometer assembly506 is used which is actually multiple spectrometers such as the multiple spectrometers516 illustrated in FIG. 33, only that portion of the spectra which has shifted due to its associated spectrometer516 need be regression fit in the above-noted manner.

Information on the presence of any wavelength shift identified by thecalibration subroutine576 also may be input to the currentplasma process module250, more specifically thepattern recognition module370 of FIG. 13 to account for this wavelength shift when comparing spectra from a current plasma processing operation with spectra in theplasma spectra directory284. Consider the situation where the200 nanometer wavelength of the current spectra from theprocessing chamber474 is being evaluated, and where a 1 nanometer wavelength shift has been identified by thecalibration assembly552 at the 200 nanometer wavelength. When looking to compare this data point with spectra in theplasma spectra directory284, the “inputting” to thepattern recognition module370 of the 1 nanometer wavelength shift for the 200 nanometer wavelength directs thepattern recognition module370 to actually look at the 201 nanometer wavelength of the corresponding spectra in this case to determine if the two intensities are within the intensity match limit of thepattern recognition module370. Therefore, the presence of the wavelength shift will not adversely affect the reliability of the results provided by the currentplasma process module250 of FIG.32.

Thecalibration assembly552 of FIG. 31, and including the calibration module574 of FIG. 40, may also be used to calibrate the output from thespectrometer assembly506 when there is an intensity shift in the spectra of the plasma in theprocessing chamber474. Intensity shifts in the spectra will typically be due to “aging” of thewindow478. “Aging” of thewindow478 as used herein means that the plasma processes which have been conducted within theprocessing chamber474 have affected theinner surface482 of thewindow478 in some manner (e.g., by forming deposits on theinner surface482, by etching theinner surface482, by a combination of forming deposits on and etching the inner surface482). Oftentimes these deposits will reduce the intensity of the light from thechamber474 which passes through thewindow478 and is directed to the spectrometer assembly516 (FIG.31). Failing to account for an intensity shift in the spectra due to thewindow478 in these cases may also adversely affect the reliability of the results provided by the currentplasma process module250. Consider the situation where thepattern recognition subroutine374 of FIG. 13 with its point-by-point analysis is employed by thepattern recognition module370 to compare the pattern of the current spectra with the spectra in the plasma spectra directory284 (FIG.9), and where the intensity “match limit” is set on 10% (a percentage difference basis). Also assume that certain deposits have formed on theinner surface482 of thewindow478 such that the intensity of light emitted through thewindow478 is reduced by 30%. In this case, even if the plasma in theprocessing chamber474 is healthy, the light that is emitted through thewindow478 will only be 70% of the actual intensity of the plasma inside thechamber474. Thepattern recognition subroutine374 will therefore indicate that the current spectra is not a “match” with any spectra in the relevant subdirectory of the plasma spectra directory284 (because its “match limit” is a 10% variation in the intensity of corresponding wavelengths in the subject spectra, and because the window deposits have reduced the intensity of light from thechamber474 by 30%), even though the plasma is in fact healthy. Therefore, a “false negative” would be reported by thepattern recognition subroutine374 and the currentplasma process module250.

The calibration module574 of FIG. 40 may also include a calibration subroutine which is generally directed toward making an adjustment to account for the above-noted type of intensity shift in the spectra of the plasma in thechamber474 as emitted through itswindow478. One embodiment of such a subroutine is illustrated in FIG. 45 in the nature of thecalibration subroutine600. No plasma exists in thechamber474 during the execution of thecalibration subroutine600 or at least when obtaining data therefore. Moreover, detection of plasma in accordance with the foregoing may automatically terminate calibration operations through thesubroutine600. Finally, thesubroutine600 may be executed on a periodic basis which is established by the operator of the facility using the wafer production system2 (e.g., daily).

A comparison is undertaken atstep604 of thecalibration subroutine600 between the spectra of the calibration light from thecalibration light source556 which is directed to thewindow478 and which may be stored in the calibrationlight spectra subdirectory310, and the spectra of that portion of the calibration light which is reflected from theinner surface482 of thewindow478 and provided to thespectrometer assembly506. Since there is no plasma in thechamber474 at this time, the light provided to thespectrometer assembly506 should be limited to that portion of the calibration light which is reflected by theinner surface482 of thewindow478. As such, thecalibration subroutine600 actually monitors/evaluates the condition of that portion of thewindow478 which may be adversely affected by plasma processes conducted in thechamber474.

The comparison of the subject spectra atstep604 of thecalibration subroutine600 and the identification of any intensity shift instep608 of thesubroutine600 may be implemented in the following manner. The spectra of the calibration light which is sent to thewindow478 may be obtained from thecalibration light subdirectory310 of FIG.9. Its intensity peaks are identified in the manner discussed above in relation to thecalibration subroutine576 of FIG. 41, as well as possibly their relative wavelength positionings. These same peaks should appear at the same wavelength (assuming that there is no wavelength shift) and “same” intensity level (taking into consideration the above-noted principles of optics) in that portion of the calibration light which is reflected by theinner surface482 of thewindow478. Since the location and intensity of one or more peaks was identified for the calibration light that was sent to thewindow478, the amount of intensity shift may be identified simply by finding these same peaks in that portion of the calibration light which is reflected by theinner surface482 of thewindow478 and noting any corresponding intensity shift. Although the intensity of the peaks alone may be sufficient to identify the corresponding peaks in that portion of the calibration light that is reflected by theinner surface482 of the window478 (e.g., by looking for the largest peak around a certain wavelength), the relative positioning of the identified peaks may be desirable/required in some cases.

The intensity shift concept addressed in relation to thecalibration subroutine600 of FIG. 45 is further addressed in relation to FIGS. 46A and 47A. One embodiment of aspectra682 of a calibration light is illustrated in FIG. 46A (e.g., a mercury lamp or other similar calibration light source with light defined by wavelengths from about 200 nanometers to about 1,000 nanometers), and this calibration light may be used by thecalibration light source556 of the calibration assembly552 (FIG. 31) and by the calibration subroutine600 (FIG. 45) to identify an intensity shift. Thespectra682 of FIG. 46A is characterized by a plurality of discrete intensity peaks686 of varying intensity. “Intensity” is plotted along the “y” axis and is expressed in “counts” which is reflective of the intensity level, while “wavelength” is plotted along the “x” axis and is in nanometers. FIG. 46A depicts the actual pattern of light which is sent to thewindow478 in the subject example and when thewindow478 has been exposed to a plurality of plasma processes (e.g., an “aged” window478). One embodiment of aspectra690 which is output by thespectrometer assembly506 is presented in FIG. 47A after the above-noted principles of optics has been accounted for and which is representative of that portion of the calibration light which is reflected from theinner surface482 of theaged window478 on theprocessing chamber474. Thespectra690 is characterized by a plurality of discrete peaks694 of varying intensity, with “intensity” again being plotted along the “y” axis and expressed in “counts” which is reflective of the intensity level, and with “wavelength” being plotted along the “x” axis in nanometers.

Thespectra690 of FIG. 47A may be compared with thespectra682 of FIG. 46A to determine if the output from thespectrometer assembly506 needs to be calibrated for an intensity shift, such as due to an agingwindow478. This comparison is again undertaken atstep604 of thecalibration subroutine600 of FIG.45. The peaks694 of thespectra690 of FIG. 47A should appear not only at the same wavelengths as the corresponding peaks686 of thespectra682 of FIG. 46A (as discussed above in relation to calibrating the output of thespectrometer assembly506 for a wavelength shift), but they should also be at the same level of intensity. For example, the peak694afrom FIG.47A and the peak686afrom FIG. 46A should be at the same intensity level, thepeak694bfrom FIG.47A and thepeak686bfrom FIG. 46A should be at the same intensity level, thepeak694cfrom FIG.47A and thepeak686cfrom FIG. 46A should be at the same intensity level, and so forth. However, this is not the case.

Peak694aof thespectra690 from FIG. 47A is at substantially the same intensity (about 9300) as its corresponding peak in thespectra682 of FIG. 46A, namely peak686a.However, peak694bof thespectra690 from FIG. 47A has an intensity of about 5,100, whereas the intensity of its corresponding peak in thespectra682 from FIG. 46A, namely peak686b,has an intensity of about 11,100. That is, there was a intensity drop of about 6000 intensity units or about a 54% drop in intensity at about the 450 nanometer wavelength region. Moreover, peak694cof thespectra690 from FIG. 47A has an intensity of about 9,600, whereas the intensity of its corresponding peak in thespectra682 from FIG. 46A, namely peak686c,has an intensity of about 12,000. That is, there was an intensity drop of about 2,400 intensity units or about a 20% drop in intensity at about the 760 nanometer wavelength region. Therefore, thewindow478 is not having the same dampening effect on the entire spectra being obtained through thewindow478. Instead, thewindow478 is having at least a first dampening effect on one part of the spectra (e.g., the 450 nm region) and a second dampening effect on another part of the spectra (e.g., the 760 nm region) which is identifiable by comparing the patterns in the noted manner. Step612 of thecalibration subroutine600 will make at least one adjustment in relation to theplasma monitoring assembly500 to account for these types of intensity shifts, and then control will be turned over to thestartup module202 of FIG. 32 bystep614 of thecalibration subroutine600 of FIG.45.

The adjustment referred to instep612 of thecalibration subroutine572 may be generally viewed as normalizing thespectra690 of FIG. 47A (calibration light reflected by theinner surface482 of the window478) to thespectra682 of FIG. 46A (calibration light sent by thecalibration light source556 to the window478). One way of “normalizing” the data is by “regression fitting.” This option may be used regardless of what type of spectrometer is implemented for thespectrometer assembly506, such that this option may be employed for both scanning-type spectrometers and solid state spectrometers. Another way of characterizing the subject adjustment is through the concept of calibration factors or gain. If there is a “uniform” intensity shift, one calibration factor or gain may be applied to the optical emissions data which is collected on the current plasma process. Multiple dampening effects identified in accordance with the foregoing may then be addressed through multiple calibration factors or gains. One or more parts of the spectra of reflected light may require the application of one calibration factor or gain thereto, while one or more parts of the subject spectra may require the application of another calibration factor or gain thereto, and so forth. Finally, thespectrometer assembly506 may be manipulated in some manner to obtain more light, although such is not as preferred as the foregoing.

Consider the example where the comparison between thespectra690 of FIG.47A and thespectra682 of FIG. 46A indicates the existence of an intensity shift. The output from thespectrometer assembly506 may be regression fit to account for the intensity shift which was identified. Alternatively, information on the presence of this intensity shift may be input to the currentplasma process module250, more specifically thepattern recognition module370 of FIG. 13, to account for this intensity shift when comparing spectra from a current plasma process with spectra in theplasma spectra directory284.

Limits may be utilized for adjustments made in relation to theplasma monitoring assembly500 by any of the calibration subroutines noted herein. For instance, in the event that the amount of calibration or gain required to address an intensity shift exceeds a first limit (or a wavelength shift for that matter), a message may be displayed to the appropriate personnel that thewindow478 has degraded or aged to the degree where the accuracy of the results provided by theplasma monitoring assembly500 may be affected to an undesirable degree. Alternatively, exceeding this first limit may in fact disable theplasma monitoring assembly500 and a corresponding indication may be provided to operations personnel. Imposing a higher limit than the noted first limit may also be used to trigger disabling of the plasma monitoring assembly500 (i.e., warn when a first limit is exceeded, and disable when a second, higher limit is exceeded).

The calibration light illustrated in FIG.46A and discussed above has a plurality of discrete intensity peaks. Therefore, only 31 data points may be evaluated by thesubroutine600 of FIG. 45 to identify the effect that thewindow478 is having on the optical emissions passing therethrough (i.e., there are only 31 intensity peaks, and the remainder is merely noise). Assumptions would have to be made as to the effect that thewindow478 is having on optical emissions at those wavelengths between these peaks. An embodiment of a calibration light which may be used by the calibration light source556 (FIG. 31) and the calibration subroutine1300 (FIG. 45) to identify an intensity shift, which alleviates the need for these types of assumptions, is presented in FIG.46B. The calibration light of FIG. 46B is of a different type than that illustrated in FIG. 46A in that the calibration light of FIG. 46B presents a continuum of intensity, while the calibration light of FIG. 46A has a plurality of discrete intensity peaks. This light may be the second type used by the secondcalibration light source556bnoted above (the first type being the calibration light of FIGS.42/46A to identify a wavelength shift). In one embodiment, the calibration light for identifying an intensity shift is a white light source defined by wavelengths which include at least the Preferred Optical Bandwidth. Therefore, thecalibration light source556 may actually include one type of light source for identifying a wavelength shift (e.g., FIGS.42/46A), and may use a different light source for identifying an intensity shift (e.g., FIG.46B).

The comparison of thespectra666 of FIG. 46B (calibration light sent to the window478) with thespectra670 of FIG. 47B (that portion of the calibration light which is reflected by theinner surface482 of thewindow478 and provided to the spectrometer assembly506) may be done in the same manner as discussed above in relation to FIGS. 46A and 47A. However, a more complete “picture” of the behavior of thewindow478 is provided by using a calibration light having a continuum of intensity since there are more data points available for comparison than in the case of a spectra which only has a plurality of intensity peaks with effectively “noise” therebetween (e.g., FIGS.42/46A). Moreover, a comparison of thespectra666 of FIG.46B and thespectra670 of FIG. 47B indicates how thewindow478 is having a different effect on different portions of the spectrum. Note how the shapes of thespectra666 and thespectra670 are generally the same between about the 200 nm and 575 nm wavelengths although with different intensities, but how the shapes of thespectra666 and thespectra670 differ between about the 575 nm and 950 nm wavelengths. Specifically, the profile of thespectra670 of FIG. 47B between about the 575 nm and 950 nm wavelengths is “flatter” than the corresponding portion of thespectra666 of FIG.46B. Therefore, thewindow478 is having one type of effect on optical emissions generally between about the 200 nm and 575 nm wavelength region, and another different effect on optical emissions generally between about the 575 nm and 900 nm wavelength region.

Another embodiment of a calibration subroutine which may be utilized by the calibration module574 of FIG. 40 is illustrated in FIG.48. Thecalibration subroutine616 is able to identify when thewindow478 is having different dampening effects on different portions of the spectral data being emitted through thewindow478. Moreover, thesubroutine616 is also able to identify when thewindow478 is completely filtering yet different portions of the spectral data being emitted through thewindow478. Preferably, thecalibration subroutine616 uses a light source which has a continuum of intensity to provide these functions, such as the light source depicted in FIG.46B.

Referring now to FIG. 48, thecalibration subroutine616 starts withstep620 which initiates a comparison between the spectra of the calibration light which is sent by thecalibration light source556 to the window478 (hereafter “reference spectra” for purposes of the subroutine616) (e.g., FIG.46B), and the spectra of that portion of the calibration light which is reflected by theinner surface482 of the window478 (hereafter “reflected spectra” for purposes of the subroutine616) (e.g., FIG.47B). Any “change” in intensity between the reference spectra and the reflected spectra is noted atstep624 of thecalibration subroutine616.Steps620 and624 may employ the same type of logic as presented bysteps580 and584 of thecalibration subroutine576 of FIG.41 andsteps604 and608 of thecalibration subroutine600 of FIG.45. The “change” in intensity referenced instep624 of thecalibration subroutine616 of FIG. 48 may be a predefined tolerance or a “match limit” as discussed above in relation to thepattern recognition module370 of FIG.13. That is, if each “point” of the reflected spectra is within the “match limit” of the corresponding “point” in the reference spectra, thecalibration subroutine616 will proceed to step628. Step628 indicates that no calibration of theplasma monitoring assembly500 is required, and that the control is passed to thestartup module202 of FIG.15.

If there is a “change” in intensity between the reflected spectra and the reference spectra (e.g., exceeding the “match limit” in the comparison of the reflected spectra and reference spectra at one or more points in the point-by-point analysis), thecalibration subroutine616 will proceed fromstep624 to step632. Step632 analyzes the manner in which the reflected spectra differs from the reference spectra in relation to the subject intensity criteria. If there was a “uniform” intensity shift of the reflected spectra in relation to the reference spectra, thecalibration subroutine616 proceeds to step636 where a single calibration factor or uniform gain is applied to the entirety of the reflected spectra (and thereby to optical emissions data obtained during a current plasma process) to “normalize” the same to the reference spectra. Control is then transferred to thestartup module202 of FIG. 15 viastep640 of thecalibration subroutine616. “Uniform” in relation to step632 of thecalibration subroutine616 need not be limited to a fixed number of intensity units (e.g., the entirety of the reflected spectra need not differ from the entirety of the reference spectra by the same amount), but instead may utilize a raw difference, a percentage difference, or a combination thereof as the “match limit” as those terms are used in relation to thepattern recognition module370 of FIG.13. For instance, any change in intensity between each wavelength in the reflected spectra which is within ±5% of the intensity of the corresponding wavelength in the reference spectra may be considered a “uniform” intensity shift for purposes ofstep632 of thecalibration subroutine616.

If the comparison between the reflected spectra and the reference spectra atstep632 of thecalibration subroutine616 indicates that there is not a “uniform” intensity shift of the reflected spectra in relation to the reference spectra, thesubroutine616 proceeds fromstep632 to step644. Generally,step644 is directed to determining if there is any evidence of complete filtering of data over any wavelength range within the reflected spectra. Under certain conditions, theinner surface482 of thewindow478 may be affected such that light emissions over a certain wavelength range will be completely blocked (e.g., thewindow478 will become opaque over this wavelength range which would be evident by an effectively horizontal line in the subject spectra at a relatively low intensity and including a zero intensity level). An example would be where thewindow478 is opaque in relation to light emissions over the wavelength range of about 300-400 nanometers, but is transparent to at least a degree over other wavelengths of light being collected and provided to the spectrometer assembly506 (FIG.31).

Any “complete filtering” of any wavelength range in the reflected spectra as discussed herein will cause thecalibration subroutine616 to proceed fromstep644 to step648. Data within the completely filtered region is ignored in any analysis subsequently provided by the currentplasma process module250 of FIG. 32 as set forth instep648 of thecalibration subroutine616. Moreover, at least two different calibration factors are applied to different portions of the remainder of the data in the reflected spectra as set forth in thissame step648, or the reflected spectra is normalized in relation to the reference spectra over that portion which is not being filtered. Again recall that it was determined that the intensity shift in the reflected spectra, in relation to the reference spectra, was “non-uniform” instep632 of thesubroutine616 such that the application of a single calibration factor to the reflected spectra would not be appropriate to “normalize” the reflected spectra in relation to the reference spectra in this case.

As noted, the analysis of any plasma process thereafter performed which is analyzed by the currentplasma process module250 will be limited to only part of the desired optical emissions data. Data within the wavelength region which is being completely filtered by thewindow478 is ignored. As such, thecalibration subroutine616 may be programmed to proceed fromstep648 to theprocess alert module428 of FIG. 14 as well. One or more alerts may be issued through theprocess alert subroutine432. For instance, it may be appropriate to apprise operations personnel at this time that thewindow478 is aging and should be replaced, that the results provided by any further executions of the currentplasma process module250 may provide inaccurate results since certain data from the current plasma process will be ignored in the comparative analysis, or both. Alternatively and if desired by the operator of the facility incorporating thewafer production system2, any exiting ofstep644 of thecalibration subroutine616 as a “yes” condition may call upon theprocess alert module428 to terminate all further plasma processing operations until thewindow478 is replaced (not shown).

Control of thecalibration subroutine616 of FIG. 48 may pass fromstep644 to step656 if there was no complete filtering of data in the reflected spectra. Calibration of the reflected spectra pursuant to step656 would then entail the application of at least two different calibration factors or a plurality of gains throughout the reflected spectra, or alternatively the normalization as discussed above. Again recall that it was determined instep632 that the intensity shift in the reflected spectra in relation to the reference spectra was non-uniform such that application of a single calibration factor would not be appropriate to “normalize” the reflected spectra to the reference spectra. In the case ofstep656, one calibration factor may be applied to the reflected spectra over the 200 nanometer to 500 nanometer wavelength region, while a different calibration factor may be applied to reflected spectra over the 501-900 nanometer wavelength region.Calibration subroutine616 then exitsstep656 to step660 where control may be transferred to, for instance, thestartup module202 of FIG.15.

Thecalibration subroutine616 of FIG. 48 may be characterized as monitoring thewindow478, and based upon these results specifying how subsequent plasma processes should be evaluated. If the monitoring of the condition of thewindow478 through thesubroutine616 determines that there is no significant dampening of optical emissions from thechamber474 through thewindow478 or there is only dampening and not any complete filtering, thecalibration subroutine616 provides that plasma monitoring operations proceed normally by comparing the “normalized” output from thespectrometer assembly506 with spectra in theplasma spectra directory284. Situations where some dampening of the optical emissions from thechamber474 through thewindow478 and where some complete filtering of spectral data over a certain wavelength range is identified by execution of thecalibration subroutine616 dictates that a different evaluation technique be utilized. Specifically, some of the desired data is ignored in the analysis.

Research Module1300—FIGS.49-51C

All of the evaluations described so far in relation to the currentplasma process module250 of FIGS. 7 and 32 have related in one manner or another to the “health” of the plasma process. Another type of evaluation which is available through the currentplasma process module250 relates to the endpoint of a plasma process or a certain portion thereof. Plasma cleans, conditioning wafer operations, and each step of a plasma recipe (whether run on aqualification wafer18 or a production wafer18) each will typically have an endpoint associated therewith (when the intended/predetermined result has been achieved, such as the complete removal of a layer of material in a plasma etch) which will typically be reflected in the optical emissions of the plasma in thechamber36 and which thereby may be used to call “endpoint.” The following will discuss endpoint in relation to a plasma step of a plasma recipe run on aproduction wafer18, although it is equally applicable to any plasma process having one or more endpoints, including without limitation those identified above.

One way of identifying endpoint through the currentplasma process module250 for a given plasma step requires that some analysis be undertaken of at least one previous execution of this same plasma step in thesame processing chamber36. In this regard, the currentplasma process module250 includes aresearch module1300 which is a submodule thereof and which is presented in FIG.49. Theresearch module1300 includes aresearch subroutine1478 which is run to identify which characteristic(s) of those optical emissions of the plasma in theprocessing chamber36 may be indicative of the endpoint of the subject plasma step, and which may then be used by theendpoint detection module1200 of FIGS. 7 and 32 to identify the occurrence of the endpoint of the subject plasma step.

Theresearch module1300 may be accessed through thestartup module202 of FIG.15 through execution ofsteps144 and148. Theresearch subroutine1478 of FIG. 49 typically is set to evaluate multiple executions of the same plasma step and identifies the optical emissions data of the plasma in thechamber36 which may be used to call endpoint. Some information about endpoint may be obtained by looking at the optical emissions data from only a single run. Theresearch subroutine1478 of FIG. 49 utilizes, but does not necessarily require, some knowledge of the plasma step, such as a time estimate of the length of time required to reach the endpoint of the plasma step. In one embodiment, this a priori knowledge may be used such that optical emissions data is obtained on the plasma step at a point in time which should include this endpoint. In this regard, a time estimate t_efor completing the subject plasma step (reaching its endpoint) is input to thesubroutine1478 by execution ofstep1480. This time estimate for reaching endpoint may be calculated based upon, for instance, knowing the etch rate of the subject process and the thickness of the layer to be etched away. Another way of identifying a time associated with the subject endpoint uses the data on the current plasma process. In this regard, each time optical emissions are obtained for the subroutine1478 (e.g., over the Preferred Optical Bandwidth and at the Preferred Data Resolution), the total intensity of these optical emissions is calculated by adding the intensity of each individual wavelength in these optical emissions. Any time there is a significant change in the above-noted sum between two adjacent-in-time optical emissions may be indicative that endpoint occurred at that time.

Step1490 of thesubroutine1478 directs that product be loaded into theprocessing chamber36 andstep1482 directs that the plasma step thereafter be initiated by the introduction of plasma into theprocessing chamber36 under the appropriate conditions to initiate the desired plasma step. Optical emissions of the plasma in theprocessing chamber36 during the plasma step are obtained for thesubroutine1478 at a current time t_cfor thesubroutine1478 through execution ofstep1484, preferably using the Preferred Optical Bandwidth at the Preferred Data Resolution. Adjustment of the “clock” of thesubroutine1478 occurs atstep1486 where the current time t_cis increased by an increment of “n.” Preferably, “n” is set at the Preferred Data Collection Time Resolution. If the new current time t_cis less than a preset value relating to the time estimate t_e,step1488 of thesubroutine1478 causes thesubroutine1478 to exitstep1488 and return to step1484 for repetition in accordance with the foregoing. Although the “preset value” referred to instep1488 may be the time estimate t_efromstep1480, it may be desirable to use a larger value to ensure that optical emissions data is obtained for the time when endpoint of the plasma step is actually reached. Another way of characterizing the foregoing is that optical emissions data should be recorded until the plasma goes off in thechamber36 to be able to identify one or more wavelengths which may be indicative of endpoint.

Steps1484,1486, and1488 of theresearch subroutine1478 are basically directed to obtaining optical emissions data of the entirety of the subject plasma step within the Preferred Data Collection Bandwidth, at the Preferred Data Resolution, and at the Preferred Data Collection Time Resolution. This data could be analyzed at this time to identify characteristics in the optical emissions data which are candidates for being an indicator of endpoint. Preferably, however, data is obtained in accordance with the foregoing on multiple executions of this same plasma step in thesame processing chamber36. This is implemented through execution of step1492 of thesubroutine1478 which directs that the foregoing be repeated for the desired amount of executions of the plasma step (at least 2 runs). Once the desired amount of data is obtained (the desired number of runs), the data is analyzed through execution ofstep1496 of theresearch subroutine1478 to identify that optical emissions data or a portion thereof which may be used as an indicator of endpoint. This analysis is currently done manually.

The analysis referred to instep1496 of theresearch subroutine1478 may be generally directed to identifying those optical emission lines (e.g., individual wavelengths) which undergo some type of a recognizable or discernible change around the time that the endpoint of the plasma step is supposed to have occurred. One type of change which would make a particular wavelength a possible candidate for being an indicator of endpoint is illustrated by a review of the optical emissions data presented in FIGS. 50A-C. Each of these figures represents a plot of intensity versus time through a time which should include the endpoint of the subject plasma step for three specific wavelengths λ₁, λ₂, and λ₃. The plots were generated from optical emissions data which was obtained from one execution of the subject plasma step on awafer18 in theprocessing chamber36. Preferably, a plot of this type is obtained for each wavelength of light which is obtained based upon the optical resolution being used in collecting the optical emissions data.

An examination of FIG. 50A for the wavelength λ₁from this first running of the plasma step reveals that there is no real discernible change at any time during the plasma step. That is, the line for wavelength λ₁is at least substantially horizontal in FIG.50A. However, the plot for the wavelength λ₂in FIG. 50B reveals two distinct changes in its emissions line. One change occurs at about the 15 second mark, while another change appears at about the 40 second mark. The plot for the wavelength λ₃of FIG. 50C also has two distinct changes in its emissions line. One change occurs at about the 20 second mark, while another change appears at about the 40 second mark. Therefore, although the wavelength λ₁can be ruled out as not being indicative of endpoint for the subject step since there is no discernible change in its entire emissions line over the relevant time period, both the wavelengths λ₂and λ₃remain candidates for being endpoint indicators because they each have at least one distinct change in their respective emission lines.

Having some knowledge of the plasma step, such as a time estimate of its endpoint, may eliminate some of the wavelengths of data obtained on the subject plasma process, some of the changes which occur in the plot for a particular wavelength, or both. For instance, if the time estimate for reaching endpoint of the subject plasma step was somewhere around 40 seconds, the changes which occurred at the 15 and 20 second time intervals in the intensity versus time plots for the wavelengths λ₂and λ₃, respectively, could be eliminated clue to their temporal spacing from the time estimate for endpoint. However, both of the wavelengths λ₂and λ₃would still be candidates for being an indicator of endpoint because there is a change in each of these plots at about the 40 second mark. No definitive determination should be made at this point without obtaining additional data in the nature of data on another running of the plasma step in thesame processing chamber36. It is important to note that all changes in the plasma are important in monitoring the plasma health and should be noted, even though they may be ignored in relation to establishing an endpoint indicator(s).

FIGS. 51A-C present the optical emissions data for the same wavelengths λ₁, λ₂, and λ₃that are presented in FIGS. 50A-C, but from another running of the same plasma step in theprocessing chamber36. The plot of the wavelength λ₁in FIG. 51A still indicates that nothing about endpoint can be derived from this wavelength λ₁. The plot for the wavelength λ₂in FIG. 51B is at least substantially the same as presented in FIG.50B. The two distinct changes in its emissions line may be due, for instance, to certain changes in the process such as the opening/closing of a valve(s). Although the plot for the wavelength λ₃in FIG. 51C has the same general pattern, the two distinct changes occur about 5 seconds later than they did in the run depicted in FIG.50C. This may be an indication that the wavelength λ₃is reflective of the endpoint of the subject plasma step where endpoint may vary in time by some acceptable time differential.

Summarizing the foregoing, wavelengths which may undergo a change which corresponds with endpoint of the subject step may be identified by comparing the plots of the various wavelengths from run to run and identifying those patterns which are generally the same from run to run, but which have some type of change. For instance, this change may be one or more of a temporal shift, an intensity shift, an expansion of the pattern, and a reduction in the pattern. These types of changes may be due to factors which vary from run to run, such as the thickness of the layer which is being addressed by the plasma step which may cause a temporal shift. Therefore, one way to to identify a particular wavelength which is indicative of endpoint is to run the same plasma step on multiple products, each having a different thickness of the layer to be etched away such that the time at which endpoint will occur will also vary. Wavelengths which have changes which may indicative of endpoint are those where the subject change (“around” endpoint) also shifts in time.

Once the wavelength(s) changes that are potentially indicative of endpoint are identified through theresearch module1300, an endpoint detection technique is still required to use this information to identify endpoint. One such techniques entails defining the pattern of that portion of the plot of intensity versus time for the subject wavelength(s) up to where the change occurred which was selected as being indicative of endpoint. Definition of this portion of the plot may be through an equation or a function (e.g., linear function, first order polynomial, second order polynomial). Endpoint may then be deemed to have been reached for subsequent executions of this same plasma step when the corresponding wavelength(s) no longer fits the equation or function. Another option is to take the first or second derivative of this function to identify the slope of the resulting line. The change in slope over time of the current optical emissions data may be plotted. Endpoint may be called when this plot deviates by more than a predetermined amount from that identified by the first or second derivative of the subject function.

There are other ways of characterizing which wavelength(s) may provide some type of indication that endpoint has been reached from the optical emissions data collected on the current plasma process. The optical emissions data obtained for theresearch subroutine1478, more specifically the plot of intensity versus time, may be reviewed to identify a peak which changes throughout the process and then reaches a steady state at about the time that endpoint is to have occurred. Moreover, these plots may be reviewed to identify a peak which remains at a steady state for the process and then begins to change at about the time that endpoint is to have occurred. A small subset of this behavior includes behavior in which a suspect wavelength or wavelengths diminish to background or a new wavelength appears at about a time where endpoint was suspected to occur.

Endpoint Detection Module1200—FIGS.52-58

The actual detection of the endpoint of a plasma process (e.g., plasma clean, conditioning wafer operation) or discrete portion thereof (e.g., one or more steps of a plasma recipe; all steps of a multi-step plasma recipe) is realized through use of the currentplasma process module250 of FIGS. 7 and 32, and more specifically through theendpoint detection module1200. Some initial comments are warranted on themodule1200. Initially, theendpoint detection module1200 may be used to identify the endpoint of a plasma process having only a single endpoint. Multi-step plasma processes having corresponding multiple endpoints may also be evaluated through theendpoint detection module1200. Two or more plasma steps of a multiple step plasma process may each have their respective endpoints identified through theendpoint detection module1200. All steps of a multiple step plasma process having endpoints also may each be identified through theendpoint detection module1200 as well.

Multiple techniques may be simultaneously employed by theendpoint detection module1200 to identify the occurrence of the endpoint(s) of a plasma process or portion thereof. One way in which this may be implemented is to utilize two or more different endpoint detection techniques and to call endpoint when any one of these techniques identifies the subject endpoint. Different in this context means that the techniques themselves are different, and not merely the data which is used by the technique. Another way in which this may be implemented is to utilize two or more endpoint detection techniques and to call endpoint when at least two of these techniques have identified the subject endpoint. This option provides a more “robust” endpoint detector by increasing the confidence that endpoint has actually been appropriately identified (e.g., statistically more confident that endpoint occurred).

Theendpoint detection module1200 may interface with theprocess alert module428 of FIG.14. For instance, when endpoint is identified through theendpoint detection module1200, information on the endpoint condition may be provided to the appropriate personnel through the “alert” function of theprocess alert subroutine432 of FIG.14. Control of the plasma process in relation to the identification of endpoint through theendpoint detection module1200 may also be affected through the “process control” function of theprocess alert subroutine432. The identification of endpoint through themodule1200 may be used to terminate the current plasma process or discrete portion thereof (e.g., a plasma step), to initiate the next aspect of the current plasma process (e.g., the next plasma step), or both through theprocess alert subroutine432 and in the above-described manner.

Preferably, the current plasma process is evaluated both in relation to its health (i.e., the plasma health in accordance with theplasma health module252 discussed above) and in relation to endpoint through theendpoint detection module1200. Although it is beneficial to monitor the health of the plasma process throughout substantially the entirety thereof (except possibly the initial portion thereof when the plasma first comes on and when the plasma is rather unstable), theendpoint detection module1200 need not be initiated at the start of the plasma process. Instead, theendpoint module1200 preferably begins its evaluation of the current plasma process to identify its endpoint(s) at an intermediate time in the process (e.g., at least ½ way into the subject process or portion thereof). However, there is nothing inherently wrong with having themodule1200 evaluate the entirety of the plasma process to identify one or more of its endpoints. It just unnecessarily “consumes” processing time/capabilities. Moreover, the plasma health evaluation and endpoint evaluation need not be executed at the same frequency or Analytical Time Resolution. The amount of time between which the health of the plasma process is checked may be greater than the time between which the plasma process is checked to identify endpoint. For instance, the optical emissions of the plasma may be checked at least at every 1 second for a plasma health evaluation, and may be checked at least at every 300 milliseconds to identify endpoint.

Another feature may be included in each of the endpoint detection subroutines to be addressed below which also relates to plasma health evaluations by theplasma health module252. If theplasma health module252 identifies either an error or unknown condition in relation to the current plasma process, this may have an effect on the operation of theendpoint detection module1200. For instance, theendpoint detection module1200 may be configured to respond to such a condition by displaying information to the appropriate personnel that endpoint will not be called because of the identification of the error or unknown condition by theplasma health module252. Moreover, if theendpoint detection module1200 is interfaced with the controls of theprocessing chamber36, identification of an error or unknown condition through theplasma health module252 may terminate the ability of theendpoint detection module1200 to affect any changes in the plasma process, to terminate a current plasma process after identifying its endpoint, to initiate the next plasma process, or both. That is, theendpoint detection module1200 may be “shut off” or disabled if the plasma health is found to be unacceptable by theplasma health module252 since endpoint information obtained under these circumstances may be unreliable.

One or more of the above-noted types of features may be incorporated in each of the endpoint detection subroutines to be addressed below. The remainder of the discussion presented herein on endpoint detection through theendpoint detection module1200 will only be in relation to a plasma step of a plasma recipe. However, it is equally applicable to endpoint detection for any type of plasma-related process.

One embodiment of an endpoint detection subroutine which may be called by theendpoint module1200 of FIGS. 7 and 32 is presented in FIG.52. Theendpoint detection subroutine1456 is used to determine when a given plasma step of a plasma recipe has affected its intended purpose or achieved the desired result. A current spectra of the plasma in theprocessing chamber36 at the current time t_cis obtained for thesubroutine1456 through execution ofstep1460. An assessment of this spectra from theprocessing chamber36 with theendpoint subdirectory316 is undertaken atstep1464 to determine if endpoint has been reached. One technique contemplated bystep1464 is to determine if this spectra from theprocessing chamber36, or at least a portion thereof, is a “match” with at least one spectra in theendpoint subdirectory316 of FIG.9. Optical emissions data over a wavelength range (e.g., the Preferred Optical Bandwidth using the Preferred Data Resolution) may be stored in theendpoint subdirectory316, such as a spectra obtained from thesame processing chamber36 during a previous execution of the same plasma step at a time when endpoint should have been reached. The pattern of the spectra of the plasma in theprocessing chamber36 will typically remain substantially constant for a time after endpoint has been reached. Analyzing the data from a previous execution of this plasma step in thesame processing chamber36 may therefore allow for identification of a plurality of spectra at about the time which endpoint should occur and which are undergoing no real significant change. These spectra may be defined by the Preferred Optical Bandwidth and Preferred Data Resolution. One or more of these “steady state” spectra may then be included in theendpoint subdirectory316 for comparison with the current spectra for purposes of identifying a “match” instep1468 of theendpoint detection subroutine1456. Utilization may be made of thepattern recognition module370 of FIG. 13 bystep1464 and1468 to see if there is a “match.”

There are other ways of identifying a “match” indicative of endpoint instep1468 of theendpoint detection subroutine1456 of FIG.52. One or more wavelengths which were identified by theresearch module1300 of FIG. 49 may have their respective plots of intensity versus time from a previous execution of the same plasma step in thesame processing chamber36 defined by an equation or function. In this case, the assessment throughstep1464 may be to plot the optical emissions data of this wavelength in the subsequent execution of the same plasma process in thesame chamber36 and to determine if this data indeed fits with this equation or function. If so, endpoint may be called when the current optical data no longer “fits” the equation or function (e.g., employing linear fit techniques, polynomial fit technique, etc) which should be at about the time where the “change” discussed above in relation to theresearch module1300 occurred. Other techniques which may be used to assess this type of current data would be to utilize a first derivative or a second derivative of this equation or function to define a linear function, such that deviations from this function may be more readily identified in some cases as noted above (i.e., when the plot of the change in slope over time of the current optical data deviates from the slope defined by the first or second derivative).

Features may be incorporated in the present invention to enhance the ability to call endpoint based upon the behavior of one or more specific wavelengths. When the wavelengths which are indicative of endpoint are identified through theresearch module1300 of FIG. 49, certain relational characteristics of these particular wavelengths may be noted as well. Consider the case where only a single wavelength has been selected as an endpoint indicator. The intensity of the peak of this wavelength may be noted to identify the same in any subsequent executions of the same plasma process. For instance, the subject wavelength may represent the largest peak within a certain wavelength region. Therefore, the subject wavelength may be found in these subsequent executions by looking for the largest peak in a certain area of the spectrum. In addition, the “position” of the peak of the subject wavelength may be identified in relation to one or more other peaks. For instance, the subject wavelength may be represented by a peak which is located between two larger peaks in a certain wavelength region. Therefore, the subject wavelength may be found in the subsequent executions by looking for this pattern in the optical emissions of the plasma.

If the assessment of the spectra of the plasma in theprocessing chamber36 at the current time t_cindicates that endpoint has not yet been reached, theendpoint detection subroutine1456 proceeds fromstep1468 tosteps1476 and1470 which may be executed in any order.Step1476 resets the “clock” of thesubroutine1456 by increasing the current time t_cby a factor of “n.” The magnitude of “n” defines the Analytical Time Resolution.Step1470 makes a determination as to whether the current plasma process has been terminated. The same types of techniques discussed above in relation to determining when the plasma is “on” in thechamber36 may be used instep1470. So long as the current plasma process has not been terminated, thesubroutine1456 returns to step1460 where another spectra is obtained for thesubroutine1456 at this new current time t_cfor a repetition of the above-described analysis. If the plasma process has been terminated, thesubroutine1456 proceeds fromstep1470 to step1466 where control of the plasma monitoring operations may be returned to, for instance, the start-upmodule202 of FIG.15. If no endpoint has been detected by the time at which the plasma process is terminated, information may be provided to personnel that endpoint was not detected and that there may have been some error or aberration in the process, even though theplasma health module252 did not necessarily identify this condition.

Theendpoint detection subroutine1456 of FIG. 52 will continue to execute in the above-noted manner untilstep1468 identifies endpoint throughsteps1464 and1468. At this time thesubroutine1456 will proceed fromstep1468 to step1474 where theprocess alert module428 of FIG. 14 is called. There, actions may be taken to apprise personnel that the endpoint of the subject plasma step has been reached (through execution ofsteps454 and458 of the process alert subroutine432), actions may be taken in relation to control of the plasma process (e.g., initiating the next plasma step or terminating the plasma recipe if the subject step is the last step of the recipe), or both. Whether through theprocess alert module428 or throughstep1472 of theendpoint detection subroutine1456, control of plasma monitoring operations may then be returned to, for instance, thestartup module202 of FIG.15.

Another embodiment of an endpoint detection subroutine which may be accessed through theendpoint module1200 is presented in FIG.53. Theendpoint detection subroutine1506 of FIG.53 initiates atstep1510 where a spectra of the plasma in theprocessing chamber36 at the current time t_cis obtained for thesubroutine1506. This spectra and a “reference” spectra are subtracted from each other atstep1514. Only the differential is important in relation to step1514. That is, it is not of particular importance whether the current spectra is subtracted from the “reference” spectra or vice versa. Preferably, both the reference spectra and the spectra on the current plasma process which is obtained are defined by the Preferred Optical Bandwidth and Preferred Data Resolution.

At least two alternatives may be implemented for the “reference” spectra fromstep1514 of theendpoint detection subroutine1506. The “reference” spectra forstep1514 may be retrieved from theendpoint subdirectory316 of FIG.9. Preferably, the spectra from theendpoint subdirectory316 which is the “reference” spectra forstep1514 of theendpoint detection subroutine1506 is a spectra which is associated with the same plasma process. That is, the spectra in theendpoint subdirectory316 which is involved in the “subtraction” ofstep1514 would be a spectra from a previous execution of the same plasma process in thesame processing chamber36 at a time in the plasma process when endpoint is at least assumed to have occurred. How this spectra is selected for inclusion in theendpoint subdirectory316 may be in accordance with the discussion presented above. Moreover, information provided by the operator, by direct communications between thewafer production system2 and the currentplasma process module250, or by thepattern recognition module370 in which identifies the recipe, recipe step, etc, are detected, on the plasma process to be run in theprocessing chamber36 may be used to select this spectra from theendpoint subdirectory316 for use in the “subtraction” associated withstep1514. Another option for the “reference” spectra for purposes ofstep1514 is a previous-in-time spectra from the same plasma process in which the current spectra is obtained at step1510 (e.g., the immediately preceding current time t_c). For instance, the spectra at the current time t_cand the spectra at the current time t_c−nmay be subtracted from each other atstep1514.

The subtraction operation involving the current spectra of the plasma in theprocessing chamber36 at the current time t_cand the “reference” spectra through execution ofstep1514 of theendpoint detection subroutine1514 generates an output which is indicative of the differential. If thesubroutine1506 determines that this differential is within a certain predetermined tolerance atstep1518, endpoint will be deemed to have been reached and the subroutine will proceed fromstep1518 to step1530 where control is transferred to theprocess alert module428 for action in accordance with the foregoing. Otherwise, thesubroutine1506 will proceed fromstep1518 to executesteps1522 and1524. The order in which steps1522 and1524 are executed is of no particular relevance.Step1522 resets the “clock” of thesubroutine1506 by increasing the current time t_cby a factor of “n.” The magnitude of “n” defines the Analytical Time Resolution.Step1524 makes a determination as to whether the current plasma process has been terminated. The same types of techniques discussed above in relation to determining when the plasma is “on” in thechamber36 may be used instep1524. So long as the current plasma process has not been terminated, thesubroutine1506 returns to step1510 where another spectra is obtained for thesubroutine1506 at this new current time t_cfor a repetition of the above-described analysis. If the plasma process has been terminated, thesubroutine1506 proceeds fromstep1524 to step1528 where control of the plasma monitoring operations may be returned to, for instance, the start-upmodule202 of FIG.15.

The “differential” referred to instep1518 of theendpoint detection subroutine1506 of FIG. 53 may be compared with a predetermined tolerance, such as baseline intensity. A raw difference basis, a percentage difference basis, or both may be used for establishing this tolerance. For example, when all of the data points involved in the “subtraction” ofstep1514 are within ±“x” intensity units of each other, thesubroutine1506 may be directed to proceed fromstep1518 to step1530. Another way of saying this is that endpoint is deemed to have been reached for purposes of theendpoint detection subroutine1506 when there are no longer any substantial peaks in the differential defined bystep1514. These types of concepts are illustrated by reference to FIGS. 54A-C,55A-C, and56A-C.

Thespectra1496aof FIG. 54A is representative of plasma in theprocessing chamber36 in the initial part (e.g., for a current time to) of a plasma step being run on product within the processing chamber36 (e.g., the spectra of FIG. 54A is the spectra at the current time t_cfromstep1510 of theendpoint detection subroutine1506 of FIG.53).Spectra1496ahas a plurality ofintensity peaks1498aat various wavelengths and at various intensities. Thespectra1500 of FIG. 54B is the “reference” spectra fromstep1514 of theendpoint detection subroutine1506 and is defined by a plurality ofpeaks1502 at various wavelengths and with various intensities. FIG. 54C is an illustration of the “difference” between thespectra1496aof FIG.54A and thespectra1500 of FIG. 54C which is presented in the nature of anoutput1504a.Note that theoutput1504ain FIG. 54C has substantial peaks therein which is indicative that endpoint has not yet been reached based upon the logic of theendpoint detection subroutine1506 of FIG.53.

Thespectra1496bof FIG. 55A is representative of plasma in theprocessing chamber36 at an intermediate time (e.g., for a current time t₃₀) in the same plasma step presented in FIG. 54A (i.e., the spectra of FIG. 55A is the spectra at the current time t_cfromstep1510 of theendpoint detection subroutine1506 of FIG.53). Thespectra1500 of FIG. 55B is again the “reference” spectra fromstep1514 of theendpoint detection subroutine1506, and FIG. 55C is an illustration of the “difference” between thespectra1496bof FIG.55A and thespectra1500 of FIG. 55B which is presented in the nature of anoutput1504b.Note that theoutput1504bin FIG. 55C still has substantial peaks therein which is indicative that endpoint has not yet been reached based upon the logic of theendpoint detection subroutine1506. However, note that the plasma step has progressed in that the size of the peaks in theoutput1504bof FIG. 55C is less than the size of the peaks from theoutput1504aof FIG.54C.

Thespectra1496cof FIG. 56A is representative of plasma in theprocessing chamber36 toward the end (e.g., for a current time t₄₅) in the same plasma step presented in FIGS. 54A and 55A (i.e., the spectra of FIG. 56A is the spectra at the current time t_cfromstep1510 of theendpoint detection subroutine1506 of FIG.53). Thespectra1500 of FIG. 56B is again the “reference” spectra fromstep1514 of theendpoint detection subroutine1506, and FIG. 56C is an illustration of the “difference” between thespectra1496cof FIG.56A and thespectra1500 of FIG. 56B which is presented in the nature of anoutput1504c.Note that theoutput1504cin FIG. 56C now has no substantial peaks therein which is indicative that endpoint has been reached based upon the logic of theendpoint detection subroutine1506. Therefore, endpoint would be called by thesubroutine1506 when the spectra of FIG. 56A is encountered for action in accordance with the foregoing.

Another embodiment of a subroutine which may be utilized by theendpoint detection module1200 of FIGS. 7 and 32 is presented in FIG.57. Generally, theendpoint detection subroutine1204 of FIG. 57 associates the endpoint of a given plasma process or a discrete portion thereof (e.g., a plasma step) with a “modal” change in the plasma within thechamber36 or a portion thereof around the time when endpoint has occurred. In many cases there is a change in impedance associated with theprocessing chamber36 at the time that endpoint is reached. This change in impedance oftentimes is reflected by a “modal” change in the plasma within theprocessing chamber36 or a portion thereof. Changes in the plasma are in turn reflected in its optical emissions. Endpoint could be called based upon this modal change alone (i.e., solely through theendpoint detection subroutine1204 of FIG.57). However, preferably this “modal” change for identifying endpoint is used in combination with another endpoint identification technique (e.g., through theendpoint detection subroutine1456 of FIG. 52, through theendpoint detection subroutine1506 of FIG.53).

Optical emissions data are obtained for theendpoint detection subroutine1204 of FIG. 57 at itsstep1208. These optical emissions data are more specifically of the plasma in theprocessing chamber36 at a time t_c1. Two other time-related variables are introduced atstep1212. A time t_c1+nis set equal to a time t_c2atstep1212. As such, the time t_c2is greater than the time t_c1by an increment of “n.” The magnitude of “n” for the subroutine defines the Analytical Time Resolution. Optical emissions of the plasma in theprocessing chamber36 are obtained for this time t_c2through execution ofstep1216 of theendpoint detection subroutine1204.

The intensities of the optical emissions from the current values for the times t_c1and t_c2are compared with each other atstep1220 of theendpoint detection subroutine1204. If the differential between the optical emissions at these two times is less than a predetermined amount (e.g., an increase or a decrease of a certain magnitude), determined through execution ofstep1224, thesubroutine1204 proceeds fromstep1224 to step1228. The time t_c2is set equal to the time t_c1atstep1228 and spectral data for a new time t_c2is obtained by theendpoint detection subroutine1204 returning to step1212 for repetition in accordance with the foregoing. In order to allow thesubroutine1204 to exit the foregoing if the plasma process is terminated, astep1240 is incorporated which will provide this function by proceeding to astep1244 where control may be returned to, for instance, the start-upmodule202 of FIG.15. Control may also be returned to the start-upmodule202 after endpoint is called in accordance with the following and through execution ofstep1236.

If the differential between the optical emissions associated with the times t_c1and t_c2is ever more than the predetermined amount associated withstep1224 of theendpoint detection subroutine1204, this may be an indication of the above-noted “modal” change in the plasma which is in turn indicative of endpoint. It is not definitively indicative of the type of “modal” change at this point in time however. Only those “modal” changes associated with the plasma which occur at about the time that endpoint is estimated to occur, which appear quickly or abruptly, and which are persistently observed in subsequent executions of the same plasma process are indicative of a change in impedance which occurs at endpoint. Therefore, it may be desirable to execute a plurality of runs for each particular plasma process in which thesubroutine1204 is used before relying upon thesubroutine1204 to call endpoint by execution of theprocess alert module428 through execution ofstep1232. Thesubroutine1204 may also be implemented to confirm or increase the confidence level that endpoint has been reached when another endpoint detection technique is being used to call endpoint as well.

The “modal” change associated with theendpoint detection subroutine1204 of FIG. 57 may be monitored in relation to one or more specific wavelengths, but by definition impacts plasma performance and is detectable throughout the optical emissions range (although some wavelengths may demonstrate a stronger effect of the modal change than others). For instance, the optical emissions associated with the differential referred to instep1224 may be in relation to a single wavelength. Moreover, the optical emissions associated with the differential referred to instep1224 may be from a plurality of wavelengths, and the “differential” which is potentially indicative of a change in impedance may be when one or more of these wavelengths demonstrates the above-noted “modal” change. Another option is to monitor for the “modal” change over a certain bandwidth. In this regard, theoutput1260 presented in FIG. 58 illustrates the change in intensity of the plasma in thechamber36 over time throughout the plasma step. More specifically, each point of theoutput1260 illustrates the differential between the optical emissions of the plasma in thechamber36 over the Preferred Optical Bandwidth and at the Preferred Data Resolution of a current time (e.g., t_c) and a preceding time (e.g., t_c−n). Preferably, each data point which defines theoutput1260 is the differential between two adjacent times where optical emissions data is obtained from thechamber36. As can be seen in theoutput1260 from FIG. 58, there is a significant change in the intensity of the plasma from about the 35 second mark and the 40 second mark. This is the type of “modal” change which may be associated with a change in impedance associated with theprocessing chamber36, which in turn theendpoint detection subroutine1204 of FIG. 57 associates with endpoint.

Wafer Distribution Module1384—FIGS.59-60

Various of the above-noted evaluations provided by the currentplasma process module250 may provide information to thewafer distribution module1384 of FIGS. 59-60 to have some type of effect on the distribution of wafers to thevarious process chambers36 of thewafer production system2. One embodiment of a subroutine which may be used by thewafer distribution module1384 is illustrated in FIG.59. Thesubroutine1388 of FIG. 55 includessteps1392,1396,1400, and1402 in which the protocol is for thewafer distribution subroutine1388 to proceed to the plasmaprocess product module252 for each ofchambers36a-d(FIG.1). Thesubroutine1388 may of course accommodate wafer production systems having different numbers ofprocessing chambers36. Monitoring of the current plasma recipes being run on product in each of theseprocessing chambers38a-dis included in the protocol of1404. Any deviation from thenormal spectra subdirectory288 of the current plasma recipe being run on product in any of thechambers36a-dis noted atstep1408. In the case where the current plasma recipe being run on product in a givenprocessing chamber36 does not “match” any plasma recipe within thenormal spectra subdirectory288, the distribution ofwafers18 to thischamber36 is suspended. This may be due to an error in the subject plasma recipe identified by the plasmaprocess product module252 of FIGS. 21-25, the identification of an unknown condition in the subject plasma recipe by the plasmaprocess product module252 of FIGS. 21-25, or the identification of a dirty chamber condition by thechamber condition module1084 of any of FIGS. 27-29. Although suspension may be implemented on the first occurrence of this type of event, thewafer distribution subroutine1388 may be structured such that there is no suspension until a predetermined number of consecutive plasma recipes have deviated from thenormal spectra subdirectory288 on a givenchamber36 or when a certain percentage of a fixed number of plasma recipes have deviated from thenormal spectra subdirectory288 on a given chamber36 (e.g., when at least one out of three runs have not matched any plasma recipe in the normal spectra subdirectory288).

Another embodiment of a subroutine which may be used by thewafer distribution module1384 of FIGS. 7 and 32 is illustrated in FIG.60. Thesubroutine1416 includessteps1420,1424,1428, and1430 in which the protocol is for thewafer distribution subroutine1416 to proceed to the plasmaprocess product module252 for each of therespective processing chambers36a-d(FIG.1). Thesubroutine1416 may of course accommodate wafer production systems having different numbers ofprocessing chambers74. Monitoring of the current plasma recipes being run on product in each of theseprocessing chambers36a-dis included in the protocol ofstep1432. The total time required to execute each plasma recipe on each of thechambers36 is also noted atstep1432.Step1432 may thereby use the plasmaprocess product module252 discussed above in relation to FIGS. 21-25, as well as theendpoint module1200 discussed above in relation to FIGS. 52-58. The distribution sequence ofwafers18 to thesechambers36 is based upon the total time required to complete each plasma recipe as noted instep1436.

The logic ofstep1436 of thewafer distribution subroutine1416 may be that a priority is given to theprocessing chamber36 which is executing the plasma recipe in the shortest amount of time. That is, the distribution sequence ofwafers18 to thechamber36 may be to utilize the “fastest” processingchamber36 the most. If the “fastest” chamber was36aand is available (i.e., no product therein), the logic ofstep1436 would instruct thewafer handling assembly44 to provide awafer18 to thischamber36aeven though another chamber (e.g.,chamber36b) is “scheduled” to receive thenext wafer18. Alternatively, the logic ofstep1436 may be that a priority is given to theprocessing chamber36 which is executing the plasma recipe over the longest period of time. Lengthening of the time required to execute a plasma recipe may indicate that thechamber36 is becoming dirty, and may be in need of “cleaning” in the near future as discussed above in relation to thechamber condition module1084 of FIGS. 27-29. Since this is the case, the operator of the facility incorporating thewafer production system2 may decide to maximize the runs on a givenchamber36 once it has been determined that it will be taken “off-line” in the near future for cleaning.

Conclusion

Theplasma monitoring assemblies174,500, and700 as described above offers numerous advantages. One advantage is that the logic of the currentplasma process module250 and its various sub-modules can be used with any type of plasma chamber and further can be used to evaluate any type of plasma process. No significant adaptation of the logic of themodule250 or any of its sub-modules is required to use themodule250 on one chamber type/design, and then to turn around and use thissame module250 on another chamber type/design. Similarly, no adaptation of themodule250 or any of its sub-modules is required to use themodule250 on one type of plasma process, and then to turn around and use thissame module250 on a plasma process which is somehow different. In this sense themodule250 is a generic plasma monitoring tool.

Another advantage of theplasma monitoring assemblies174,500,700 is that the data which is used to evaluate the plasma processes conducted within a processing chamber is taken from this very same processing chamber when running the very same type of plasma process. In this sense when theassemblies174,500, and700 are installed on a particular processing chamber, theassemblies174,500,700 become specific to the chamber. The fact that a given plasma recipe will behave in one way on one chamber and another way on another chamber is of no significance to theplasma monitoring assemblies174,500,700. Eachassembly174,500,700 adapts to the idiosyncrasies of its associated chamber and learns about the chamber through the running of actual plasma processes on the chamber to build up theplasma spectra directory284 for that particular chamber. Again, each chamber will have its ownplasma spectra directory284 or the like in that all information used to evaluation a given plasma process on a given plasma chamber will have been obtained from this very same processing chamber.

Significant information on the current plasma process is made available pursuant to the current plasmaprocess product module250 and its various submodules. Themodule250 has the ability to: determine if the current plasma process is proceeding like one or more previous runnings of this same plasma process on this same chamber; determine if the endpoint of a given plasma step has been reached; determine if the endpoint of each plasma step in a plasma recipe has been reached; identify the current plasma recipe to operations personnel; identify the current plasma step to operations personnel; determine when a chamber should be cleaned; and to determine when each of a plasma cleaning, and conditioning wafer operation may be terminated. How this information is used will typically be left up to the discretion of the operator of the facility incorporating thewafer production system2. For instance, the currentplasma process module250 may simply be used a source of information on the current plasma process. The currentplasma process module250 may also take a more active roll in the operation of thewafer production system2 by integrating themodule250 with process controls for thesystem2, such as to automate control of plasma process operations based upon information provided by themodule250. Various combinations may also be employed in that any sub-module may be “programmed” to be for information only, for process control only, or for both. Flexibility to deal with these types of issues is provided by the modular construction of the currentplasma process module250 where themodule250 and each of its sub-modules are again preferably stored on a computer-readable data storage medium (e.g., computer-readable memory such as one or more hard disks, floppy disks, zip-drive disks, CDs). Therefore, changes to the structure of the currentplasma process module250 may be readily implemented.

The foregoing description of the present invention has been presented for purposes of illustration and description. Furthermore, the description is not intended to limit the invention to the form disclosed herein. Consequently, variations and modifications commensurate with the above teachings, and skill and knowledge of the relevant art, are within the scope of the present invention. The embodiments described hereinabove are further intended to explain best modes known of practicing the invention and to enable others skilled in the art to utilize the invention in such, or other embodiments and with various modifications required by the particular application(s) or use(s) of the present invention. It is intended that the appended claims be construed to include alternative embodiments to the extent permitted by the prior art.

Claims (22)

What is claimed is:

1. A method for preparing a processing chamber for running a plasma recipe on product when loaded in said chamber, said method comprising the steps of:

plasma cleaning an interior of said processing chamber with a plasma, wherein said plasma cleaning step is executed when said processing chamber is empty;

obtaining optical emissions of said plasma within said processing chamber at least a plurality of times during said plasma cleaning step, wherein said obtaining step comprises obtaining an optical emissions segment of said plasma in said processing chamber at each of said plurality of times, wherein each said optical emissions segment comprises optical emissions which includes at least wavelengths from about 250 nanometers to about 1,000 nanometers, inclusive, which defines a first wavelength range, and at a first wavelength resolution throughout said first wavelength range, wherein said first wavelength resolution is defined as a wavelength spacing between each adjacent pair of wavelengths throughout said first wavelength range, and wherein said first wavelength resolution is no more than about 1 nanometer;

comparing a pattern of each of a plurality of said optical emissions segments from said obtaining step with a first standard pattern which is stored on a computer-readable storage medium;

monitoring said comparing step for a first condition, wherein said first condition is when said pattern of at least one of said optical emissions segments from said comparing step matches said first standard pattern and which is equated with an endpoint of said plasma cleaning step; and

terminating said plasma cleaning step if said monitoring step identifies said first condition.

2. A method, as claimed in claim1, wherein:

said plasma cleaning step comprises removing deposits which have adhered to interior surfaces of said processing chamber during the running of plasma recipes on product within said processing chamber before execution of said plasma cleaning step.

3. A method, as claimed in claim1, wherein:

said plasma cleaning step comprises removing residuals from interior surfaces of said processing chamber, wherein said residuals exist from a wet clean which was conducted within said interior of said processing chamber before execution of said plasma cleaning step.

4. A method, as claimed in claim1, wherein:

each of said patterns from said comparing step and said first standard pattern are a plot of intensity versus wavelength at a specific point in time.

5. A method, as claimed in claim1, wherein:

said first standard pattern was obtained from a previous-in-time plasma cleaning operation conducted in the same said processing chamber.

6. A method, as claimed in claim1, further comprising the steps of:

opening said processing chamber before said plasma cleaning step;

replacing at least one component within said interior of said processing chamber after said opening step and before said plasma cleaning step; and

sealing said processing chamber after said replacing step, wherein said plasma cleaning step is executed only after execution of said sealing step.

7. A method, as claimed in claim1, further comprising the steps of:

opening said processing chamber before said plasma cleaning step;

applying at least one type of solvent to at least a portion of interior surfaces of said processing chamber before initiating said plasma cleaning step; and

sealing said processing chamber after said applying step, wherein said plasma cleaning step is executed only after execution of said sealing step.

8. A method, as claimed in claim1, further comprising the step of:

monitoring an amount of time spent on said plasma cleaning step; and

discontinuing said plasma cleaning step if said monitoring an amount of time step identifies a second condition, wherein said second condition is when said amount of time spent on said plasma cleaning step is at least equal to a predetermined maximum time limit.

9. A method, as claimed in claim8, further comprising the step of:

initiating a wet clean of said interior of said processing chamber after any execution of said discontinuing step.

10. A method, as claimed in claim1, wherein:

said first standard pattern is defined by optical emissions of wavelengths within said first wavelength range and at said first wavelength resolution throughout said first wavelength range.

11. A method, as claimed in claim10, wherein:

said monitoring said comparing step comprises determining a differential between said pattern of each of said plurality of said optical emissions segments from said comparing step and said first standard pattern, wherein said first condition exists when any said differential is free from any substantial peaks.

12. A method, as claimed in claim10, wherein:

said monitoring said comparing step comprises determining a differential between said pattern of each of said plurality of said optical emissions segments from said comparing step and said first standard pattern, wherein said first condition exists when any said differential is within a predetermined amount of a baseline intensity.

13. A method for preparing a processing chamber for running a plasma recipe on product when loaded in said chamber, said method comprising the steps of:

plasma cleaning an interior of said processing chamber with a plasma, wherein said plasma cleaning step is executed when said processing chamber is empty;

obtaining optical emissions of said plasma within said processing chamber at least a plurality the times during said plasma cleaning step, wherein said obtaining step comprises obtaining an optical emissions segment of said plasma in said processing chamber at each of said plurality of times, wherein each said optical emissions segment comprises optical emissions which includes at least wavelengths from about 250 nanometers to about 1,000 nanometers inclusive, which defines a first wavelength range, and at a first wavelength resolution throughout said first wavelength range, wherein said first wavelength resolution is defined as a wavelength spacing between each adjacent pair of wavelengths throughout said first wavelength range, and wherein said first wavelength resolution is no more than about 1 nanometer;

determining at each of a plurality of times a differential between a pattern of two of said optical emissions segments from different times of said obtaining step;

monitoring said determining step for a first condition, wherein said first condition is when said any said differential from said determining step is at least substantially a flat line without any substantial peaks; and

terminating said plasma cleaning step if said monitoring step identifies said first condition.

14. A method, as claimed in claim13, wherein:

said plasma cleaning step comprises removing deposits which have adhered to interior surfaces of said processing chamber during the running of plasma recipes on product within said processing chamber before execution of said plasma cleaning step.

15. A method, as claimed in claim13, wherein:

said plasma cleaning step comprises removing residuals from interior surfaces of said processing chamber, wherein said residuals exist from a wet cleaning operation which was conducted within said interior of said processing chamber before execution of said plasma cleaning step.

16. A method, as claimed in claim13, wherein:

said determining step comprises determining an amount of said differential between said pattern of each of said plurality of said optical emissions segments from said obtaining step with said pattern of said optical emissions segment from an immediately preceding time of said obtaining step.

17. A method, as claimed in claim13, wherein:

said determining step defines a time rate of change of said optical emissions within said processing chamber during said plasma cleaning step.

18. A method, as claimed in claim13, further comprising the step of:

monitoring an amount of time spent on said plasma cleaning step; and

discontinuing said plasma cleaning step if said monitoring an amount of time step identifies a second condition, wherein said second condition is when said amount of time spent on said plasma cleaning step is at least equal to a predetermined maximum time limit.

19. A method, as claimed in claim13, further comprising the steps of:

opening said processing chamber before said plasma cleaning step;

replacing at least one component within said interior of said processing chamber after said opening step and before said plasma cleaning step; and

sealing said processing chamber after said replacing step, wherein said plasma cleaning step is executed only after execution of said sealing step.

20. A method, as claimed in claim13, further comprising the steps of:

opening said processing chamber before said plasma cleaning step;

applying at least one type of solvent to at least a portion of interior surfaces of said processing chamber before initiating said plasma cleaning step; and

sealing said processing chamber after said applying step, wherein said plasma cleaning step is executed only after execution of said sealing step.

21. A method, as claimed in claim13, wherein:

said determining step comprises sequentially determining a differential of said pattern of each said optical emissions segment from said obtaining step with said pattern of said optical emissions segment from an immediately preceding time of said obtaining step.

22. A method, as claimed in claim13, wherein:

each said pattern is a plot of intensity versus wavelength at a specific point in time.

Images